Full text of "Condensed Handbook Of Composition Input (book)"

See other formats

Condensed

Handbook

Composition

Input

Frank J. Romano
President

Graphic Arts Marketing Associates
Salem, New Hampshire

Produced in conjunction with

National Composition Association

Section of

Printing Industries of America, Inc.

GRAPHIC COMMUNICATIONS CENTER, 1730 NORTH LYNN STREET, ARLINGTON, VIRGINIA 22209 TELEPHONE: 703/527-6000 TELEX: 89-2738

CABLE ADDRESS: PRINTINAM, ARLINGTON, VA.

This book is dedicated to the memory of Arthur L.
Koop, Jr., Manager of Advertising and Sales
Promotion for Mergenthaler Linotype Company.
Thanks Art.

The bulk of typesetting for this book was done on Photon 713-5
photo typesetters using the Warlock Random Access Composition Entry
system for input. The card medium (covered in Chapter 5) permitted the same
kind of correction flexibility that would have been available in hot metal slugs.
The author wishes to thank Warlock, its president, Ernest Griesbach, and
especially its pretty and efficient keyboard operators. It is also with great
appreciation that thanks are extended to the many manufacturers that
provided information and assistance; and also to Roy Mochi, and to Don
Goldman.

Table of contents

1. In the beginning

2. Telling a typesetter what to do

3. Keyboard to tape systems

4. Input media and coding

5. Punched card input systems

6. Video display terminal (VDT) systems
. Optical character recognition (OCR) systems

8. What a computer does

9. Word processing

10. Keyboard arrangements

Page 1
Page 9
Page 35
Page 55
Page 63
Page 71
Page 85
Page 107
Page 117
Page 139

Early in 1926, Walter W. Morey suggested that equip¬
ment could be developed for automatic operation of
linecasting machines from one or more remote points.
This suggestion was discussed with Frank E. Gannett,
owner of a chain of newspapers in the eastern U.S.

Later that year, Gannett and Morey discussed this
matter with Sterling Morton, Howard Krum, and Ed¬
ward Kleinschmidt, owners of what was then the
Morkrum-Kleinschmidt Corporation, which manufac¬
tured most of the wire communication equipment in
this country. The name of this corporation was later
changed to the Teletype Corporation. After several
conferences, it was agreed that Teletype would
manufacture the equipment developed by Morey.

In 1927, a Teletypesetter Perforator was ready - based
on the principles used in the six-unit perforator then
manufactured for communication purposes. The
equipment was tried at the McCarty Typesetting
Company in Chicago, and during the first day fifty
five lines were set covering the Lindbergh flight to
Paris.

By 1928, the equipment was considered sufficiently
well developed for prelininary unveiling to the field.
Accordingly, a demonstration was arranged on the
top floor of the Rochester Times Union Building (see
photo to right).

During 1929, a separate corporation was organized
under the name of the Teletypesetter Corporation to
handle this type of business, and from then to 1930 an
experimental installation was made at the Evanston
(Illinois) News Index.

Manufacture of equipment was curtailed during the
war. By 1951 the press associations began tran¬
smitting justified tape to newspapers and after one
year about 400 daily newspapers were using this ser¬
vice.

The use of Teletypesetter grew at a rapid pace. In
1958 Fairchild Graphic Equipment purchased the
business of Teletypesetter and in 1964 reported that
over 1200 daily newspapers, 400 weekly newspapers
and 350 commercial shops were using its tape equip¬
ment.

In use today, there are about 15,000 Perforators and
12,000 Operating Units. In 1972 Fairchild was sold
to the Varityper division of Addressograph
Multigraph. Below are two new versions of what
began way back in 1926 when one man had an idea -
an idea and an accomplishment that has to be con¬
sidered one of the milestones in the progress of the
graphic arts industry.

1. In the beginning

Over four hundred years passed before Gutenberg’s movable type moved by
machine rather than by hand. Even as late as the last decade of the
Nineteenth Century and the tail-end of the Industrial Revolution a
newspaper compositor set type almost the same way a Fifteenth Century
printer did. From a large case in front of him he selected the necessary
letters, numbers and other characters to form the line of type and then filled
out the line with spaces to the required width. He did this over and over
again for every line. A good worker could zip along at about 40 of today’s
newspaper lines an hour.

The development of the Linotype is a unique story all its own and this is not
the place to retell it. Suffice it to say that Ottmar Mergenthaler, after several
false starts, decided not to use foundry (or handset) type but rather to cast an
entire line from matrices. One of his major problems was the automatic
justification of the line of type which was solved by the sliding wedge or
spaceband patent he obtained from J. W. Schuckers. Typesetters, on the
average, set type by machine six times faster than by hand. Employers, hard-
nosed businessmen that they were even then, sought to increase this
profitability by utilizing unskilled operators, preferably women. However,
the unions and experience showed that re-trained hand compositors were
more appropriate. An active Fern-Lib movement in 1900 could have caused

i |

a different kind of revolution in the composing room.

In 1887 Rolbert E. Lanston unveiled the Monotype. Although constructed
differently in many ways from the Linotype, its major difference involved
the non-direct operation of the machine. A perforated tape was prepared on
one unit, and then run on another to cast the characters. Corrections were
made without re-setting an entire line as with the Linotype. It is interesting
to note, with 20-20 hindsight, that tape typesetting was born at the same time
the Linotype was.

In 1928 Walter Morey, a Monotype operator, invented the Teletypesetter.
He experimented with tape as the typesetting medium to run a specially
adapted Linotype, applying the principles of the Monotype to the dominant
typesetting device of the day.

Composition Input 1

Typists, with minimal training, could produce tapes, which when fed into the
Linotypes, produced metal slugs as though an operator was at the keyboard.
In 1951 the Associated Press announced its TTS service and a year later
United Press did the same. News now came into a newspaper as a tape which
could activate a Linotype to produce type automatically. Billboard, a weekly
newspaper, that year signed the first TTS union contract, outlining plans for
perforation of tapes in New York for transmission by wire to its midwestern
composing room. By 1953 almost 700 daily newspapers utilized TTS
equipment, totalling about 1,400 keyboards for “punching” tape and about
1,500 units for feeding tape to the Linotypes.

Teletypesetter Corporation promoted its system by saying that a beginning
typist could produce 400 or more lines of type an hour. In 1958 Fairchild
Graphic Equipment, Inc. purchased the assets of Teletypesetter and
introduced new models which offered greater type mixing capability and
other productivity features. In 1971 it was reported that over 15,000 TTS
keyboards were built, with many still in use.

Since Linotype operators required retraining to use a TTS keyboard, union
schools were established. The ITU even developed the Brewer Keyboard,
invented by Claire N. Brewer, which was essentially a Linotype Keyboard
designed to fit over a TTS keyboard. The operator now punched tape on a
familiar keyboard arrangement.

Operating a TTS keyboard was more fatiguing than typing. There were
nineteen additional keys over the normal typewriter set; the operator had to
move his eyes from copy to pointer regularly to monitor justification, plus
scan the tape for possible errors. In spite of this, the major advantage of tape
operation was not at the keyboard (at this point in time) but in the efficiency
of Linotype operation. Three Linotypes running from tape could produce as
much type as seven Linotypists. By 1959, almost 1,200 newspapers had
installed TTS equipment.

The greater portion of this editorial page was composed on the first
commercial Linotype placed in a Newspaper Plant. It revolutionized com¬
posing room methods and made possible the great Newspapers of today.

N. Y. Tribune Editorial Page, July 3, 1886

A woodcut of Mergenthaler demonstrating his new
machine. Whitelaw Reid (right) of the New York

Tribune named the Linotype. The first newspaper page set on the Linotype.

2 Composition Input

To convert the typesetting process from hand to machine a number of
developments were necessary. The most important, and basic principle on
which the linecaster rests, was that of producing lines of type to the same
width. This was solved by the spaceband, a wedge-shaped device that
expanded the space between words and forced each line to its maximum
width. The spaceband was (and still is) a mechanical device with minimum
and maximum expansion values; whereas in future systems the function of
the spaceband would be (and was) imitated arithmetically.

The Monotype System introduced tape operation long before the advent of
TTS. Consisting of two units, the keyboard and the caster, Monotype
computes what the linotype expanded. Like any successful system,
Monotype rests on a basic principle. In this case it is the Monotype unit
system. First of all, it is necessary to understand the steps that lead to
justification of a line of type:

1. The width value of every character is added, plus a minimum value for
each word space.

2. The resulting total from (1) is subtracted from the pre-determined
line length.

3. The number of word spaces is counted and this total is divided into the
remainder left from (2).

4. The amount of space determined by (3) is inserted at each word space.

5. The line now equals the required measure.

In formula form:

Total line length less
Total of characters and spaces

Individual word space

. .. .— = amount needed to

XT , r , space line out

Number ol word spaces

to measure

Thus it is extremely important that the width value of every character is
known. This is the purpose of the unit system, a method that will be re¬
incarnated in future tape operation.

Monotype matrices were designed with a maximum value of 18 units. This is
a relative measurement, not an actual width value. The “W”, “M” and other
characters are allotted the maximum of 18 units. Here is a chart indicating
characters segregated according to a unit system. Designers may vary
characters to fit a different unit value; thus, not all letters always occupy the
same unit value, hut all must fit into one of the established unit categories.

Character allotment in an eighteen unit proportional
spacing system.

Unit

width Characters

6 il.,'-

7 jft

8 I

9 rsz01 234567890$; : !

10 c e o

abdghnpquvxykJS

12 Z

13 CTL

14 ABFOP&QV

15 wDEGRUXYHKN

M m M W %

: i»cr j nit is exactly 1/18th of the width of the capital letter

Composition Input 3

As the Monotype Keyboard operator punched the special tape for activation
of the caster, a mechanical wheel “added” the unit values of each character
and space in a line. Word spaces per line were also counted and indication
made of the need for an end-of-line decision. A scale informed the operator
which keys were to be struck to obtain the correct word space width. The
resultant tape was then fed into the caster tail-end first so that the spaces
would be cast first.

Impetus for the development of tape-run linecasting came in the 1930’s with
the growth of newspaper groups and chains. The ability to share tapes
among several papers, since certain types of articles of news were produced
word for word in many different locations, seemed to be an answer to the
problems of composition productivity. The existence of the nationwide
Teletype communications system added fuel to the tape-selling fire. In 1933
Walter Morey introduced Teletypesetter and the ability to input copy a
single time on tape and transmit it to various locations by telegraphic
methods.

Tape perforation was a whole new ballgame. It was unlike linecasting; its
keyboard was a standard “QWERTY” one type rather than the Linotype’s
“ETAOIN,” and instead of monitoring the assembly of matrices, the
operator monitored a width accumulation pointer to make the necessary
end-of-line decisions. Since computational functions as to the number of
characters, number of spacebands and their minimum width values were
performed at the keyboard, the tape replaced the linecasting operator but
continued to perform all of the linecaster’s functions.

TTS was a transmission system in its original form. Tape generated at a
central location was transmitted by wire to another location. There a
duplicate tape was produced on a Reperforator. Linecasters were equipped
with the Operating Units to read the tape and Adaptors to replace the
keyboard.

The TTS Perforator had a standard typewriter layout with the exception of
the SHIFT key which did not work like a standard typewriter SHIFT key.
To access a capital letter the operator hit SHIFT, then the character, and
then UNSHIFT in order to return to the lower case. An important part of
the Perforator was the counting mechanism, which accumulated (added)
character and space values. TTS required a unit system somewhat similar to
the Monotype Linecasting System. Matrices were made according to values
established by dividing the EM space into eighteen parts. Eleven groups of
width values, starting at 6 units (the narrowest) and going to 18 units (the
widest), were used to divide up a font.

The TTS Perforator produced no hard copy (that is, no typewritten end
product). It was among the first “blind” keyboards and its only function was
to create proper input tapes for a linecaster. The operator observed the scale
in front of him and made end-of-line decisions (hyphenation, etc.) when the
pointer entered the justification zone. Other keys on the perforator
controlled linecaster machine functions.

Since the unit system restricted TTS to newspapers, another version of the
4 Composition Input

perforator was developed to count non-unit matrices. The Multiface
Perforator, as it was called, could handle those typefaces designed on a 32
unit system. The EM was divided into 32 units with 5/32 of the EM as the
narrowest character. There were 28 groups of width values.

Because all typefaces vary in the widths of their characters, especially in
fonts of different point sizes, TTS had to solve the problem of mixing. The
answer came in the form of a width encoder, called a Counting Magazine by
TTS, which recorded the width values for all characters in a particular
typeface. From unit cut, to non-unit cut, to mixing, Teletypesetter evolved
into a broadbased tape system for automatic linecaster operation.

Photographic typesetting enters the picture

In 1968 the direct input concept rose from the ashes
in the form of Compugraphic’s 7200 display
phototypesetter. It produced type from 14 to 72 point
in any of four typefaces all under direct operator con¬
trol.

There are differing opinions as to the birthday, and the father, of
photographic typesetting. Practically, as well as diplomatically, there were
several of both. In 1944 an ITT engineer in France named Rene A. Higonnet
visited an offset printing plant in Lyon and observed the difficulty of
adapting hot metal composition to the photo-lithographic process. Recalling
a technical paper on the use of flash tubes to study high speed mechanisms
such as propellers, he discussed an application of this principle with a friend,
Louis Moyroud. By 1945 they had put together a simple machine to
demonstrate that a flash was fast enough to project on a plate a clear image
of a character from a revolving disc. The second problem, line justification,
was solved by April 1946 in a unit that composed lines on film from a
selection of 82 characters. The main components of the machine were:

In May, 1946 Higonnet visited the U.S. and, attempting to interest
investors, found his way to William W. Garth, Jr., President of Lithomat
Corporation developers of lithographic plate materials. An agreement was
reached and a second version was demonstrated in July 1948. This unit
offered 88 characters, improved speed, and one of the first electric
typewriters, an Electromatic. A variable escapement mechanism permitted a
selection of 18 character widths. In 1949 the machine was shown to the
industry.

a. A manual typewriter with permutation bars and
contacts to give each character a unique binary code.

b. A memory register to record the codes on a
revolving drum. Solenoid-controlled pins could be
set for either of two positions.

c. A counter-justifier to accumulate character widths
and determine word space increments after each
line had been typed.

d. Control circuitry composed of telephone relays
(ITT, remember?)

e. A photographic unit made up of a film carriage

and a continuously rotating disc with 82 clear characters
on a black background (a photo positive).

Composition Input 5

This is the Photon 200B. It was the first true
photographic typesetter to be introduced com¬
mercially. Like the Linotype it was controlled direc¬
tly by the operator, who, sitting at the keyboard,
could set sixteen typefaces in twelve type sizes in a
variety of typographic formats up to 72 points.
Whether composition should be produced directly or
indirectly (off-line tape) is still an area of concern to
those who set type.

In September, 1950, at the Chicago-held Graphic Arts Exposition, Intertype
unveiled the Fotosetter, a machine that composed type on film or
photosensitive paper, instead of hot metal slugs. To the naked eye the
Fotosetter was a linecaster, except that the pot of molten lead was replaced
by camera equipment. It contained 114 instead of 90 keys and used a matrix
very similar in outward appearance to the casting matrix, again with the
important exception of a photo-negative character embedded in its side.
Characters could be enlarged or reduced from 6 to 36 point in a number of
sizes in-between from the 12 point Fotomat at the turn of a dial.
Complicated mixing of sizes and typefaces was greatly simplified over the
hot process where a separate magazine was necessary for each size in each
typeface. Mergenthaler, at this exhibition, showed an aborted version of the
Linotype that used ebonite matrices (the line was set and the characters
photographed). Linofilm, as it was called, went back to the drawing boards.

The Fotosetter and the Higonnet-Moyroud machine continued the direct
operation concept of the linecaster.

Meanwhile back at the H-M unit, development continued. It was decided
that the machine would operate from a standard typewriter keyboard, and
that all controls should be at the keyboard position. Other decisions:
justification should be from one typing, line lengths up to 42 picas
(Linotypes used a standard 30 picas; 42 pica units were more money), line
length should be under dial control; typefaces should be mixed in the same
line, as well as type sizes, lines should be able to be “Killed”, and all
quadding functions should be automatic. They were far from achieving these
goals. So more money was raised and Lithomat changed its name to Photon.

By 1955 ten machines had been installed. Time for development of the
expanded unit allowed time for other competing units to reach fruition. The
Linofilm from Mergenthaler (now a true photographic device using a grid
for master images) and the Monophoto from Monotype were announced.
The Photon unit was way ahead in terms of capability, offering sixteen fonts
in twelve different sizes. In late 1955 further improvements were made and
the 200 series machines were introduced.

6 Composition Input

The year 1956 saw the following inventory of industry phototypesetting
devices:

Fotosetter

Photon

Monophoto

Linofilm

ATF

250 since 1950
62 since 1950
6 since 1955
2 since 1955

31 since 1954

about $33,000 each
about $55,000 each
about $35,000 each
about $57,000 for one
keyboard , one photo unit
about $14,000 each

The American Type Founders unit incorporated the Flexowriter keyboard
to prepare tape to drive a photographic output unit. In 1951 the Friden
Company introduced the Justowriter, a two-unit direct impression
typesetting system. An operator typed his copy on one electric typewriter
(Recorder) perforating a 7-level paper tape, at the same time. Codes on the
tape for characters and justification were needed for the output unit
(Reproducer) which typed justified lines. ATF replaced the output unit with
a true photographic, instead of strike-on, device. The Monophoto unit
retained the tape principle used in its casting units. Linofilm also advocated
the separation of keyboarding and photographing.

Aware of the tremendous growth in the use of punched-tape from about 1950
to 1960, Photon decided to develop a tape capability into the 200 series
machine. Two models were planned; one to accept TTS tape (the 510 series)
and one to accept Monotype tape (the 520 series). Of course, users were
restricted in accessing all capabilities of the Photon photo unit because of the
inherent limitations of the TTS and Monotype code structures. To solve this
problem and better meet the competition of the multiple keyboards to one
photo unit concept, a new series was begun. The 540 series used the photo
unit and keyboard from the 200 series but separated them. In December
1962, 12 photo units and 5 keyboards were shipped. The keyboards
perforated 8-channel paper tape.

By 1964 new photographic typesetters were being readied. Between then and
1967, the Linofilm Quick, Fairchild Phototextsetters, Photon 713 series and
Intertype Fototronic were introduced. Most accepted 6-level TTS paper
tape, by now the dominant input medium. In 1968 Compugraphic
Corporation introduced the forerunners of a phototypesetting line priced
below all others in the industry. TTS tape was the input. Many (and that
means many) other phototypesetters were introduced between 1968 and the
present. A survey of them and their capabilities is included later in this book.

As can be seen, typesetting and input evolved along parallel, but separate
paths.

Composition Input 7

Mergenthaler’s Linofilm was introduced shortly after
the Photon 200B. It separated the functions of
keyboarding and output and was the first true
phototypesetter to demonstrate the principle. Two to
three keyboard operators could keep one photo unit
busy.

The Elektron was Mergenthaler’s new breed of
Linotype in the early sixties. It was designed
specifically for tape operation and could zip along at
fifteen newspaper lines per minute. The Intertype
Monarch was also designed specifically for tape, and,
in fact, was introduced and sold without a keyboard.

position Input

2. Telling a typesetter what to do

There was a time when typesetting meant only the process of assembling
individual pieces of type by hand. Gutenberg printed his Bible from movable
type in the middle of the Fifteenth Century in this way. The first major
advance in typesetting came centuries later with the invention of the casting
typesetters. Another major advance in typesetting occurred effectively
within the last five years with the development of photographic typesetters.

Typesetting cannot be easily discussed without reference to hand typesetting
and to the operations of Linotype and Monotype machines. This is so
because many terms used in typesetting have no literal significance except
with respect to the earlier processes — since it was with these earlier
processes that the terms originated.

It is the purpose of this section to present some basic facts about type
composition. All of the important terms will be introduced.

An understanding of typesetting requires having at least an elementary
knowledge of typography. The most basic element of this knowledge is in
knowing what the various printing characters are called.

Printing characters

In the English alphabet there are twenty-six letters with which words are
formed. Each letter appears in several kinds of printing characters: upper
case letters, lower case letters, small capitals, initial letters, ligatures, and
logotypes .

Upper case letters are called capital letters since they are used for satisfying
the rules of grammar which requires that the first letter in every sentence be
a capital and that proper names be capitalized.

The terms upper case and lower case are derived from the arrangement of
cases used by printers for holding type. Capital letters are distributed in the
upper case; hence, the term upper case letters.

Composition Input 9

Small capitals are similar to upper case letters but are designed somewhat
differently in order to form entirely capitalized words having a pleasing
appearance. The height of small capitals is always somewhat reduced from
that of upper case letters.

Initial letters are ornamental or large capital letters used to embellish
composition.

A ligature is formed by combining two or more letters into a unique design.
Early printers, following the practices of the scribes, had many more
ligatures than is now customary. In book composition and in many
periodicals, but not newspapers, the f-ligatures are common.

Ligatures for the dipthongs ae and pe are familiar printing characters for
both upper and lower case letters and for small capitals.

A logotype , defined with respect to handset type and Monotype, is a
combination of printing characters cast from a common matrix. The
simplest kind of logotype is a combination of letters spaced in a way that is
not possible with ordinary type. For example, the italic letters Y and e
spaced according to the physical requirements of ordinary type result in an
unattractive gap; the same letters, with improved spacing, look much better
as a logotype. Because there is no inherent limitation in photographic
typesetting in achieving any desired spacing, the provision for generating
logotypes is simplified. Modern phototypesetters utilize a “kern” command
to “back up” and tuck characters together.

Many logotypes are distinctive combinations of letters and other printing
characters. Ligatures can be included within the definition of a logotype.

Accented letters are part of many alphabets: French, German, Italian,
Spanish, any foreign language.

Figure is the term used by printers when referring to the numerals 0, 1,2, 3,
4, 5, 6, 7, 8, and 9. There are two distinct classes of figures. Modern figures
are designed to have a uniform height and sit on the base line of the alphabet
to which it belongs. Old style figures are designed so that some figures
descend below the base line and the figures do not all have the same height.

Fractions are closely associated with figures. An extensive variety of
fractions is available in type.

Points, or the marks of punctuation include the following:

, Comma
; Semi-colon
: Colon
— Dash
Hyphen

4 Beginning quotation marks
’ Ending quotation marks

10 Composition Input

[ Beginning brackets
] Ending brackets
(Beginning parenthesis
. Period

? Question Mark
! Exclamation Point
) Ending parenthesis

The quotation marks are used simply or doubly. The apostrophe is of course
identical as a printing character to a single ending quotation mark.

The ampersand is a mark which represents the word and. There are common
and italic forms of the ampersand.

The marks of reference include the following:

* Asterisk
t Dagger
$ Double Dagger
§ Double-S

The paragraph mark can take many forms.

The following commercial and monetary signs are familiar printing
characters:

Per Cent

Pound

Number

c/o

Care Of

Dollar

Sterling

Superior and inferior characters include both letters and figures and are used
in mathematical work. Superior figures are used to denote references in
scholarly texts.

Special characters are used in printed matter for a wide variety of purposes:
ornamental devices, arrows, astronomical marks, mathematical symbols,
ecclesiastical signs, musical notations, bullets, ballot boxes, etc.

Horizontal and vertical rules are used to form tables. A careful examination
will show that each vertical line is composed from a series of short rules.

Leaders are dots used in many types of composition, such as telephone
directory listings.

An EM leader has two dots; an EN leader has one dot.

The term typeface originated with hand typesetting. A typeface is the flat

Composition Input 11

surface at the top of a piece of type; when subsequently inked and pressed
against paper, an impression of the printing character is obtained. The face
is thus a reverse image of the printing character.

The term face has broad significance in that it is used in describing styles of

type. Thus one can generalize about bold face types or become specific and
refer to Caslon Old Face.

A typeface may have a number of features which have an effect on spacing
and on how well characters are reproduced. A stem is a thick vertical line
used in forming a letter; a hair line is a thin one. Some letters consist only of
straight lines and others, like the p consist also of loops or swells. A counter
is a space surrounded by lines. A serif is a short cross-line at the extremity of
a letter and is one of the most important characteristics of type style. Sans
serif characters have no serifs.

The middle parts of letters used in the same composition must all sit on a
common base line. The part of a letter that drops below the base line is called
a descender, the part that rises above the middle part is called an ascender.
The entire vertical distance in which a type face is to be located is called the
body height. The corresponding horizontal distance is called the width.

The face of a piece of hand-set type is produced by casting. A matrix for the
face is inserted into a mold. Molten metal is poured into the mold; the chief
ingredient of the alloy is usually lead. Hand-set type is often referred to as
foundry type because it is produced in foundries.

The term matrix is directly applicable to both Linotype and Monotype
casting operation. A Monotype matrix is used to cast at a given time a single
piece of type quite similar to foundry type. A Linotype matrix is also called a
mat. Linotype matrices are assembled as shown in the figure at right and a
complete line of type is then cast in the Linotype mold. The resulting line of
type is called a slug.

Type is produced in photographic typesetting by projecting images onto
photographic film or paper. Images are created by transmitting light
through clear areas on an otherwise opaque film, drum, or disc. The
characters formed by the clear areas may be called font characters. Font
characters may also be generated using cathode ray tubes.

Linotype lead slugs. The ability to access only a por¬
tion of typeset material for correction is one of the at¬
tributes of hot metal composition that photographic
methods have been hard put to emulate.

Font characters are analogous to the matrices of the type casting processes
with an important exception. The correspondence between a matrix and the
resulting typeface is one-to-one; however, the typeface produced from a font
character need not be the same size as the font character. In the
Compugraphic 7200 Phototypesetter, for example, a single font character
can produce typefaces in eight different sizes by selecting from eight
projection lenses.

Type measurements

The height, width, and spacing of type is measured in points and picas. There
are three principal point systems in use today. The American-British System

12 Composition Input

is in use in English speaking countries. The point in this system is 0.01383
inch; twelve points makes a pica. The Didot System is in use in France and in
most countries of Continental Europe — but not in Belgium. The basic unit
of the Didot System is called the corps or point ; it has a value of 0.01483
inch. Twelve corps equal a Cicero. In the Mediaan System of Belgium, the
corps or point has a value of 0.01324 inch; twelve of these equals a Mediaan
and/or Cicero.

Two measurements determine the size of an individual piece of type.
Referring back, these are the body height and width. For foundry and
Monotype, the measurements can be made on an actual piece of type. Type
produced photographically, however, does not have any apparent boundries;
but the boundaries, nonetheless, are implicit. The artwork from which
photographic characters is made is drawn within an area having an outline
corresponding to the borders of an equivalent piece of foundry type.

When referring to a complete alphabet rather than to an individual piece of
type, the alphabet size is defined by body height and set. Body height is
always measured in points.

Point size always refers to the body height. Thus, 10-point Scotch Roman
specifies a Scotch Roman type having a 10-point body height.

The width of an individual piece of foundry type is determined by the design
of the printing character. For example, a 7-point lower case letter i would
ordinarily be much narrower than a 7-point upper case B in the same
alphabet.

The width allotted a Linotype face is the same as the width of the Linotype
matrix. Since Linotype matrices are made from brass, the term brass width
is often used to denote the width of a Linotype character.

The set of an alphabet is a relative measure of the width of the entire
alphabet. The em space is divided into a number of relative units (RU). The
width in points of the em is the set of the alphabet. Every character in the
alphabet is designed so that its width is an integral number of relative units.
The actual width of a character in points is then

I-ARACTERS

Matrices and spacebands assembled and ready for

casting.

RU in character

x Set

RU in em

where set is in points. In most typesetters the alphabets are designed with an
em having 18 relative units; Justowriter alphabets are designed with an em
having 5 relative units.

For foundry type, an em is a square space having a width equal to the body
height.

The body height of type produced on a phototypesetter is determined by the
size of the font character and by the magnification of the lens used to project

Composition Input 13

PROJECTION LENS

FONT CHARACTER

an image of the font character. The Compugraphic 2900 and 4900 Text
Phototypesetters have two lenses so that the type size is controlled by the size
of the font character (same size lens) or twice size (2X lens). In the CG 7200
Headline Phototypesetter the font character is always 10 point. By means of
eight projection lenses, type from 14 points to 72 points can be produced
from the same 10 point font characters.

When a character has been photographed, the width stepping motor will
take as many steps as there have been relative units assigned to the
character. The motion of the width stepping motor is modified by the set
gears in such a way that the lens carriage will move so that the optical axis of
the system will shift an amount equal to the width of the type. Different sets
are provided for by changing set gears.

LENS

CARRI

WIDTH STEPPING MOTOR

Font & font strips

A font of type consists of an assortment of printing characters in a size and
style. Thus, ten point Electra is the description of a particular font. The
exact assortment of type in a font will vary somewhat depending on the
requirements of composition.

A Standard Roman font includes upper and lower case letters, small
capitals, figures, marks of punctuation and reference. Some fonts may
contain ligatures or fractions, and others may not.

A font of foundry type is stored in the printers’ cases. There is a separate
partition for each character. A character for which there is no partition is
called a pi character. When two different fonts of foundry type are used in
the same composition, the compositor simply goes from one pair of printers’
cases to another to collect the pieces of type required.

In Linotype operation, two or more fonts are available, depending on the
capabilities of the particular machine in consideration. This is accomplished
by two operations: the rail shift and the magazine shift.

Some Linotype machines are equipped with two magazines — an upper
magazine and a lower magazine. Some Linotypes have as many as four
magazines; these are usually designated as MAG-1, MAG-2, etc. Each
magazine stores an assortment of ninety different matrices. A matrix that is
not stored in a magazine is called a pi mat or pi matrix and is said to run pi.
Referring back to the figure of the Linotype matrix, it is seen that a Linotype
matrix contains two different faces. When a matrix is assembled, either one
or the other of the faces is aligned in the casting position. This is
accomplished by the rail shift. A character is thus assembled in the lower rail
position, or upper rail position. By means of the rail shift, companion faces
can be composed in the same line.

The concept of upper and lower rails, and upper and lower magazines is
applied to the identification of the style choices available. A font strip for a
CG 2900 Series Phototypesetter has a single track of characters. The first
half of the font strip contains the lower rail style, and the second half the
upper rail. The font drum of a CG 7200 Phototypesetter can mount two

A two-font film strip. Upper half for lower rail; lower
half for upper rail.

LOWER

RAIL

FONT

UPPER

RAIL

FONT

LOWER

UPPER

RAIL

LOWER

MAGAZINE

UPPER

MAGAZINE

A four font film strip. Now the upper half is the lower
magazine and the outside row the upper rail.

14 Composition Input

UPPER MAGAZINE
FONT STRIP

Film strips are mounted on drums with position
nomenclature the same as that used on the linecaster.

The Photon glass matrix disc.

LOWER MAGAZINE
FONT STRIP

separate font strips. One font strip is mounted in the lower magazine (LM)
position, and the other in the upper magazine (UM) position. In addition,
each font strip has two tracks, one of which is designated as being in the
lower rail, and the other in the upper.

- .recasting magazine. It has 90 channels, each for a
raaque character - in one point size.

The term composition when applied to printed matter, means the selection
of type by size and style, and its arrangement. Even a cursory survey of
newspapers, books, circulars, calling cards, posters, and pamphlets will give
some indication of the great variety possible with printed matter.

The size and style of type and its spacing are selected to satisfy readability
and appearance. The general appearance of a column or page of type will
create a distinctive light or dark appearance. This appearance is referred to
as the color of type. In photographic type compositions color is determined
not only by the selection of the face, but by the photographic process. The
control of photographic density is thus very important in achieving good
typography.

Straight matter is simple justified text.

Except for the paragraph indention of the first line, every line is flush with
the left margin ; except for the last line, every line is flush with the right
margin. The length of a line is measured in picas and points. Type not flush
with the left or right margin is ragged.

i'E'R RAIL CHARACTER

ER RAIL CHARACTER

A linecasting matrix.

A line of foundry type is justified by filling the line with words; or, following
the rules of hyphenation, by ending the line with part of a word and a
hyphen.

To fill out a line so that it is flush with the right margin, spaces are inserted
between words called interword spaces. In Linotype composition, the
interword spaces are filled out by inserting space bands between words.
Space bands are double wedge devices that expand in thickness when
mechanically acted upon.

Composition Input 15

In some cases it is necessary to add spaces between the letters of a word - a
process known as letterspacing.

Justification in computer-controlled photographic typesetting is effected by
calculation. The space required to fill a line is divided into the number of
interword spaces and, if necessary, one or more words are letterspaced.

Sometimes the spaces in a number of lines of type cause noticeable white-
connected areas called rivers.

The various spaces used in composing a line have names related to their
relative thickness. An em space in photocomposition is a space having a
width based on the full number of units used in designing the type. Thus for
an 18 unit alphabet, the em has eighteen relative units. The actual width, of
course, depends on the set of the type as previously explained. In foundry
type, the em is a square of the type body. A square space is also called a
quad. The term quadding applies to filling out a line with space. An en space
has half the thickness of an em space. The space used in letterspacing is often
called a thin space or a hair space.

. The spacebands are pushed up, or “expand” to fill the

When composing foundry type, the spacing of characters is determined by word s P aces -

the width of type and the use of additional spaces. In photocomposition,

spacing is determined by the distance one photographic image is offset from

another. This offset is measured from the vertical reference line of one

character to the vertical reference of the following one. The term escapement

is often applied to this width because some early phototypesetters used

escapement mechanisms for obtaining character displacement.

In some composition it is desirable to place type closer together than is
possible in the case of foundry type if kerning were not used. Kerning allows
part of a character to extend into the body of an adjacent character. Kerning
is, of course, easily accomplished in photographic typesetting.

The terms flush left, flush right, and center are used to describe unjustified
lines flush with the left margin, the right margin and centered. Display
composition, the type of composition used in advertisements, utilizes flush
left, flush right, and centered lines. The term Quad is synonymous with
Flush).

The Linotron 505 was introduced by Mergenthaler in
late 1967. It utilizes a cathode ray tube (CRT) for
composition and originally required input via a com¬
puter. The present version, 505C, has its own com¬
puter and can thus accept unjustified tape. Mergen¬
thaler has a special keyboard designed specifically for
the 505.

16 Composition Input

The space between lines of type is controlled in foundry type by the body
height of the type and by the addition of leading between lines. Leading is
strips of lead, hence the term. In photographic type composition, the spacing
is determined by the amount by which the base line of a line of type is
displaced from the baseline of the preceding line. The term leading when
applied to photocomposition means the displacement of lines of type.
Leading is usually measured in points.

Justification

Here’s how a line of type is justified in a “computer” system or stand-alone
typesetter with “computer” capability:

All typesetting starts with initialization. One must define the parameters and
format of the material to be set in type: point size, type style, leading, and
line measure. This is also called “dressing” the typesetter.

S*t Solid

As characters are keyed (input directly or input via tape) they enter the
computer’s or typesetter’s logic under the same kind of scrutiny that dance
hall girls come on stage. The width value of every character is deducted from
a counter on which the total line measure has been recorded.

Characters are, of course, read at speeds of a few to a few hundred a second.
The typesetter continues to deduct width values and also counts the

Certain letter combinations may be examined in the more sophisticated
computer system, such as an “ff ’ combination. Here the computer may be
programmed to replace this letter set with a ligature for more aesthetic
typography. Also, this combination is noted in case a line ending must be
made. Characters continue to be deducted.

By now the end of the line is approaching. Width deduction is approaching
the “justification” zone. This zone is determined by a calculation that counts
the number of word spaces and establishes minimum and maximum
expansion values. If a word space cannot expand, the line is too tight; if it
can expand too far, the line is too loose.

18 Composition Input

All characters that may possibly fit on a line are “in”. However, if the line is
too tight, a letter or several letters must be removed. There are three
alternatives: drop the word to the next line, or hyphenate it and drop certain
characters to the next line, or force justify the line.

First, hyphenation is tried. There are two basic ways: by rules of logic,
grammatical constants that will work in a majority of cases, or dictionary, a
look-up system where words are compared to a separate word list with
acceptable break points indicated.

One character at a time is dropped until a good hyphenation point is
reached.

If it isn’t, the machine will attempt to keep all characters on the line with
word spaces shrunk to their absolute minimum width.

We have now input all our characters and determined our line ending point.
If our measure was set at 30 picas and all characters total 22 picas, the eight
pica difference is divided into the number of word spaces. In this example
there are eight word spaces, so each space is one pica. If the word spaces
were too wide, thin or hair spaces would be inserted equally between all
characters. This is called letterspacing.

Measure
Total character width
Total interword space
Number of spaces

divide

30 picas
22 picas

picas

1 pica per interword space

The final line is now photographed, the same total width (measure) as every
other line. All left and right margins line up. This is justification, one of the
most important functions performed in typesetting.

And by the way, the entire sequence described, although it may vary among
typesetters or computers, takes something less than one quarter second to
occur.

second

20 Composition Input

How to drive a typesetter

The preceeding information has laid the groundwork for an understanding of
what a typesetter does. To review
A typesetter:

a. Produces various typographic characters

b. Produces various sizes of characters

c. Produces various styles of characters

d. Produces various line lengths of characters

e. Produces various placements of characters
(e.g.: quadding, leading)

Thus an input device must provide the methodology to access the specific
characters and capabilities of which the typesetter is capable. The
illustration on this page is a list of all the flashable characters in one font of a
102-character phototypesetting font. Thus the input device must have
sufficient keytop designations to allow the operator to locate and set these
characters. The input device must also have the ability to access all or some
of the following typesetter commands:

ili» nnfsi£Bnation. . .
"• fl ; mw - nrfiguration

CASE CHAR

(Shift) ACTER

5 4 3 2 1 0

0 1 2 3 4 5 6 7

III

*11

■11

■ ill.;

Ilf'

!i"

0 0 0 11 0 0

0 0 0 11 1 0

110 0 1 0 0

110 0 1 1 0

0 1110 0 0

0 1110 1 0

0 10 0 1 0 0

0 10 0 1 1 0

0 0 0 0 1 0 0

0 0 0 0 1 1 0

0 110 1 0 0

0 1 I 0 I 1 0

110 10 0 0

110 10 1 0

10 10 0 0 0

10 10 0 1 0

0 0 110 0 0

0 0 110 1 0

0 10 11 0 0

0 10 11 1 0

0 I 1 I 1 0 0

0 1111 1 0

10 0 10 0 0

I 0 0 I 0 1 0

1110 0 0 0

1110 0 I 0

0 110 0 0 0

0 110 0 1 0

110 0 0 0 0

110 0 0 1 0

10 110 0 0

10 110 1 0

10 111 0 0

10 111 1 0

2 .

8 .

10 .
11 .
12 .
13.

Quad Right
Quad Left
Quad Center
Insert Space
Insert Leader
Insert Rule
Tab Set
Tab

Back or Foward One Unit
Back or Foward One Point
One-half Point Line Space
One Point Line Space
Line Measure Set

CASE CHAR-

(Shift) ACTER

14. Line Space Set

15. Point Size Set

16. Type Style Set

17. Space Only

18. Flash Only

19. Carriage Reset

20. End Line

21. Super Shift

22. Discretionary Flyphen

23. Cancel Word

24. Cancel Line

25. Cancel Character

CASE CHAR-
(Shift) ACTER

TTS Bit Configuration. 5 4 3 2 1 0 TTS Bit Configuration. ..

ACM Code Configuration. 0 1 2 3 4 5 6 7 ACM Code Configuration

CHARACTER CHARACTER

5 4 3 2 1 0

0 1 2 3 4 5 6 7

Y .

z .

Z .

Base Line Rule (BLR)

Star.

Asterisk.

Check.

FRACTIONS

Three-Quarters.

One-Quarter.

One-Half.

One-third.

NUMERICS

1 .

Superior 1.

2 .

Superior 2.

3 .

Superior 3.

4 .

Superior 4.

0 1

5 .

. 1

0 1

Superior 5.

. 1

0 1

6 .

. 1

0 1

Superior 6.

. 1

0 1

7 .

. 0

0 1

Superior 7.

. 0

0 1

8 .

. 0

0 1

Superior 8.

. 0

0 1

9 .

. 1

0 1

Superior 9.

. 1

0 I

0 .

. 1

0 1

Superior 0.

. 1

0 1

PUNCTUATION

0 1

Semi Colon.

. 1

0 1

Colon.

. 1

0 1

Close Paren.

. 1

0 1

Open Paren.

. 1

0 1

En Bullet.

. 1

0 1

. 1

0 1

Hyphen.

. 0

Dash.

. 0

1 1

Percent.

. 1

1 1

Superior $.

. 1

1 1

Slash.

. I

1 1

Superior Cent.

. 1

Query.

. 1

Ampersand.

. 1

1 1

Comma.

. 0

1 1

Comma.

. 0

1 1

Period.

. 1

1 1

Superior Period.

. 1

1 1

Close Quote.

. 0

1 1

Open Quote.

. 0

1 1

Dollar ($).

. 0

1 1

Exclamation.

. 0

Characters that can actually be photographed include
these above (ACM 9000).

Composition Input 21

These tell the typesetter where and how to put the characters on the output
medium. It is probably evident by now that an input keyboard, attempting to
access all characters and functions via single or double keystrokes, would
require a bewildering array of keys. Here is a photo and layout of
Compugraphic’s ACM 9000 Keyboard. Note how it attempts to deal with
single keystroke command of type style and size, two ever-changing areas of
display composition. Single strokes here mean faster keyboarding. However,
to access type style and size changes for the ACM 9000 via a standard 6-level
keyboard requires a minimum of three keystrokes. To call in type style on
Row 1, the operator keys: SUPERSHIFT, “F” (for face), “1” (for Row 1).
Let us review why multiple keystrokes are needed.

The ACM 9000 direct entry keyboard.

The standard typewriter keyboard layout has 44 character keys (including
SPACE); each key usually controlling two characters, which are selected via
the SHIFT key. There are 26 alphabetic keys (for capitals and lower case
letters), 10 numerical keys (for figures and special characters), 5 punctuation
keys (but only 8 unique characters out of 10 possible positions, since the
period and comma are repeated in the SHIFT position), and 2 extra keys for
fractions and rule line. 84 unique characters from 88 character positions.
The SHIFT and RETURN keys are function keys since they control the
way the machine operates.

22 Composition Input

The teletype has only 32 keys to access 64 character or function positions.
There are only capital characters.

The Linotype Keyboard layout has a total of 90 keys (plus the ELEVATE)
to access the 90 channels of the magazine. Character matrices in each
channel contain a duplexed pair of characters; that is, there are two separate
positions, called UPPER RAIL and LOWER RAIL, both of which are the
same width value. Thus the linotype has 90 keys to access 180 characters.
This represents two complete alphabets in one type size and style, duplexing
a typeface with its companion italic or boldface version. Note that the layout
is arranged with the more frequently used characters more prominently
positioned. Shown is keyboard Diagram 12.

Along came the TTS Standard Perforator. It combined the standard
typewriter layout with the function controls necessary to run the linotype.
Compare the character set from both layouts. Missing are the small caps,
since they were not used for newspaper work. The Linotype keyboard
Number 282 the newspaper version (282 f included fractions). Unlike the
typewriter, the TTS keyboard clearly differentiated between SHIFT and
UNSHIFT. These were unique codes. The RUB OUT was a correction key:
the tape was backed up to the incorrect code and when struck, RUB OUT
inserted holes in all channels. UR and LR accessed the Upper Rail and
Lower Rail position of the matrix. The BELL key signaled the receiving end
of a TTS transmission that a tape was coming in. Other versions of the TTS
keyboard were available to set non-unit cut typefaces, and to mix other faces
and sizes by commanding UM and LM (Upper Magazine and Lower
Magazine) changes. Keytops were also changed to represent characters
running in the Linotype channels that deviated from standard layouts.

As phototypesetting came more into use the number of characters per font
increased. Sizes were changed by positioning lenses instead of entirely new
fonts. More typefaces were available at one time. And electronically-
controlled photographic typesetting provided more capability and thus more
commands were necessary to access them. Intent on utilizing existing
keyboards to drive some of the new typesetters, users were forced to strike
more than one key to change a size or face or even flash one of the extra
characters over the normal keyboard set.

Subsequent keyboards increased the character and command set on the
keyboard and, in some cases, provided multi-code keys. These keys
perforated two or three codes in sequence with one keystroke. Two trends
have gone foward simultaneously since then: 1. to adapt phototypesetters to
run from the existing reservoir of keyboards, and 2. to design keyboards
specifically to run phototypesetters. The result has not been one of logical
order and method.

This is Compugraphic’s 2961. It was the first
phototypesetter ever priced under $10,000. For the
price a user received a two-font typesetter that could
accept justified and unjustified six level tape. A hard
wired computer, based somewhat on CG’s Justape,
provided hyphenation and justification.

Composition Input 23

Phototypesetting is in its third generation. First generation devices utilized
operating principles from hot metal linecasting machines but used
photography in place of actual metal casting.

Second generation phototypesetters use photomechanical methods to select
and expose character images on photographic material. These machines
were developed in the 1950’s and have undergone continual refinement ever
since. Third generation devices are fully electronic machines based on
cathode ray tube (CRT) technology. They were developed around 1965 and
are still undergoing significant changes.

Second generation photopypesetters produce typeset copy from master
characters stored photographically within the machine. The major
subsystems used to produce the copy include input, character selection,
image output control, and interline separation. Although the basic principles
used by all manufacturers are similar, numerous variations in design
complexity and configuration present a vast array of alternative methods for
setting copy.

Phototypesetters use codes from an input source to activate machine
responses that produce typeset copy. The codes identify parameters that
control the operation including: (1) characters to be typeset, (2) desired fonts
and point sizes, (3) interword spacing for line justification, and (4) desired
interline spacing.

Linofilm Quick was supposed to be a low cost
phototypesetter when it made its debut in 1964. It was
priced around twenty thousand dollars and competed
successfully with the sixty grand units then
dominating the marketplace. The Quick could only
accept justified six level tape. An optional reader was
available to accept Justowriter tapes but as far as is
known only one was ever sold.

Input sources for second generation devices are: (1) direct keyboard entry,
(2) keyboard-controlled perforated paper tape, and (3) computer-controlled
paper or magnetic tape. Early systems used entry by keyboards directly
interfaced with the machine and were designed for use by skilled operators
who understood the function codes required to typeset copy and who could
divide the copy into justified lines. This method was slow because of end-of-
line justification decisions and extra coding strokes had to be made
manually, (Photon 200 series).

As faster, and more flexible phototypesetters became available, the use of
keyboard perforating units for paper tape became more popular. These units
changed keyboard text entry into an off-line operation and made the use of
multiple keyboards for a single phototypesetter possible. This method also
allowed the typesetter to operate at its maximum speed rather than that of
the keyboard operator.

24 Composition Input

Photon’s 713 series of phototypesetters incorporated
the necessary logic to produce hyphenless
justification. With switches on the control unit, func¬
tion commands could be over-ridden in order to re¬
set copy without rekeyboarding.

Perforated paper input tapes may also be generated by general purpose
computers. These computers normally provide the flexibility and speed to
perform composition work on a variety of large, complex typesetting jobs as
well as other business applications. General purpose computers are capable
of taking information from computer-generated data banks, such as
telephone directories or parts catalogs, and, under the guidance of an
appropriate program, prepare it for phototypesetting without the step of
keyboard entry. Elimination of this step greatly enhances the efficiency and
accuracy of the total system, as keyboard entry has historically been a
bottleneck in the photocomposition process.

An additional input source available with some phototypesetters is
computer-generated magnetic tape. Magnetic tape carries the same coded
input data as paper tape and permits much higher entry speeds with
improved handling characteristics.

Character selection consists of a light source, master character set, and
control logic to synchronize the light source and character set. The light
source is of high intensity and may be continuous or may operate
stroboscopically with a flash duration of one or two microseconds
(millionths of a second). The light source normally used is a xenon flash
lamp or flash tube. For larger point sizes the lamp is flashed twice (multi-
flash) to assure that the film is fully exposed for uniform character blackness
(density).

Composition Input 2 5

A basic phototypesetting system: light source, lens,
prism or mirror and photosensitive material.

MASTER CHARACTER SET

LIGHT SOURCE

PRISM

Master character sets are transparent negatives (clear characters on opaque
backgrounds) which supply character images in the typeface or font desired.
Master character sets are stored on: (1) rotating discs, (2) rotating drums
with interchangeable film strips, (3) rotating turrets, or (4) stationary grids.
In all cases, master character sets are interchangeable to permit insertion of
desired fonts not already in the machine. Some phototypesetters offer
multiple character arrays with up to five discs or four grids.

OUTPUT MEDIA

The process of character selection begins by coded information from the
input source being read into the machine’s control logic. As the logic unit
recognizes the character and font code, it determines where the desired font
is located in relation to the light source. In machines with single character set
arrays, all available fonts are located in front of the light source. Multiple
character set arrays may find the desired font elsewhere and move it into
position before the light source. Simultaneously, the logic unit is selecting
the proper light source, when more than one is available, and positioning the
apertures, mirrors, and lenses used to isolate the desired character.

As the proper character image approaches the light source, most systems use
a series of timing marks on the master character set or timing pulses to
synchronize action between the light source and the character image. On
some machines the timing mechanism will cause the character to be
projected to stop in front of the light source. However, on most machines the
extremely fast stroboscopic light source is sufficient to obtain high quality
output by simply flashing at the exact moment the character image is in
front of the light source. Once the proper character is selected and its image
is optically initiated, image output control begins.

Image output controls and positions the character image on the output
media. Components of the subsystem include an optical projection
mechanism to focus, direct, and magnify the projected light rays and a
spacing mechanism to accurately provide intercharacter and interword
spacing.

26 Composition Input

_ J

Ll _

1 L

» _

J 1 _

L J

1 1

_l -L

- J -

1 L «

X J

L l.

—

J Li

- - 1

1 _

- J

J_1

.1 X.

L L

J _ .1

LlL

1 -J

' X

L L L

1 X L

L I

X «

L_1 » X

_»

LlL

JL _J.

Ll 1

JL -1

_»

*-1

J _i J

J L_!_

i J

-1

J J JL

L 1_l~

L -

1 J

_i L

*-1.1

J_‘

GRID

Optical projection mechanisms vary significantly among phototypesetters.
Rotating character sets normally carry several fonts around the rotating
surface. The stroboscopic light flash illuminates a complete row of
characters rather than isolating individual characters. A series of apertures,
mirrors, or lenses moves into a unique set of positions in response to function
codes on the input tape. The desired character image is thus guided into the
optical path while all others are deflected away. As the number of fonts on
an individual rotating character set increases so does the complexity of the
optical system required to isolate and project individual characters.

Stationary grid systems are the most complicated systems of all for optical
character isolation. Since character images on the grids are stationary and
not rotating past the light source, a continuous light source is used. Light is
distributed evenly over the grid by condenser lenses specifically designed for
this purpose. The effect is illumination of an entire matrix of character
images rather than a single row. The light rays from each character are
formed into a beam of virtually parallel rays by a series of lenses called a
collimator assembly. The rays then enter a series of eight pairs of optical
wedges designed to isolate the desired character image. Directed by codes on
the input tape, the wedges move to a unique set of positions for that
character. The refractive angles on the wedges guide light rays from the
appropriate character along the optical path and deflect all others away. A
decollimator assembly is used to converge the parallel light rays and
restructure the character on the output media (Linofilm Quick series).

In typesetting, point size is a measure of type height. It measures the
maximum height a font requires when adding together the maximum
distance reached by characters extending both above and below the base line.
A 72-point font requires one full inch to accommodate all characters within
it. Normal text type ranges from 5 to 12 points and most newspaper and
book headings are less than 18 points. Characters in point sizes over 18
points are normally considered as display characters.

Photographic matrices come in varying formats: disc,

drum, film strip, turret and grid. Font grids installed on the Mergenthaler V-I-P.

Composition Input 27

All phototypesetters have available a range of character point sizes, called
the point size range, but differ significantly in the number of specific settings
available and the method for changing point size. Systems vary from
manual insertion to highly automatic selection of numerous settings.

Systems that require manual changing of point size permit by far the most
straightforward and inexpensive design of the image output control
subsystem. In most of the low-priced text-oriented machines no mechanism
to control point size is required, as manual changing of the master character
set and the set-width gears is required. In several machines, manual
adjustment of point size is accomplished by the setting of levers to
correspond to the desired point size. Movement of the levers causes lenses in
the main optical system to change position and vary magnification of the
output.

In the Fototronic system, variations in magnification are accomplished
automatically by movement of two lenses. Directed by function codes on the
input tape, the movable lenses are driven along the optical axis by motors
until positioned in the appropriatelocations for the desired magnification.

The Harris-Intertype Fototronic. Input to this high
quality phototypesetter is via special counting
keyboard or computer.

The most commonly used method of obtaining different point sizes is a
multi-lens turret. The turret normally holds eight or more lenses which
provide a range of magnification levels. Each magnification level relates to a
specific point size. The approprite lens is rotated into position in response to
a coded command from the input tape. Lens turrets provide good
typographic flexibility with reasonable photomechanical simplicity.

28 Composition Input

There are numerous methods by which proper intercharacter or interword
spacing can be achieved. Two basic components are involved, the
mechanism that physically sets and advances character images across the
output media and the logic system which controls the amount of advance
and synchronizes it with the stroboscopic light source.

A lens carriage which moves laterally across the output media to locate the
character properly along the typeset line is the most inexpensive method
available. After each character is set, the lens carriage advances a distance
equal to the appropriate character width and is ready for the next character
image. Master character widths are either permanently hardwired into the
circuitry of the phototypesetter or are available through changeable width
circuits called width plugs. The character widths are then applied to the
carriage advance through a pair of set gears that must be changed for each
change in point size.

Another simple method for placing characters along the line is a positioning
mirror that rotates to deflect the image the desired amount. Character focus
is maintained by a focusing lens that moves in synchronization with the
mirror. Rotation of the positioning mirror is controlled by a second optical
system that generates electrical pulses to indicate the amount of rotation
occurring. Rotation will stop when image deflection equals the appropriate
character width. Perforated program tapes are used to load character widths
into the machine memory in coded form. A new character width tape must
be read into memory for each change in point size.

A third technique for correctly positioning characters along the line of copy
involves a carriage that advances the light source and master character set
literally in relation to the output media. The carriage advance is determined
by the character width contained in the device’s memory. Loading of the
memory from special coded tapes before typesetting begins is again used.

After the 200B, what...tape of course. Photon brought
out additional units that only ran from tape. The 513
and 560, shown, were part of a series of units. The
560 was the first computer slave.

Immediately adjacent to the master character set are two movable apertures,
one with two horizontal slits and one with a single vertical slit. Light passes
through these apertures only at the intersection of the slits. Directed by the
machine’s control logic, the apertures are positioned so that light rays from
only the desired character image pass through the intersection. However,
due to control circuitry built into the machine, the projection of a specific
character’s image need not originate in precisely the same location each time
the character is used. Timing of the flash and movement of the apertures can
be controlled to vary the optical path between the displacement limits.
Character width, interword spacing, and intercharacter spacing are all
considered during movement of the optical path. Only after maximum
displacement of the optical path has occurred will the carriage advance.

Interline separation or leading is simply the distance separating sequential
lines of typeset copy. After each line is set, the end-of-line function code
from the input source triggers a mechanism that advances the film or paper
the proper distance. Leading information is often entered through a manual
setting on inexpensive phototypesetters but is generally entered as a function
code on larger, more versatile machines.

Third generation CRT phototypesetters produce typeset copy by means of
electronic selection and exposure of character images on the face of a CRT.
The major subsystems comprising a CRT phototypesetter include date
input, character storage, character generation, and output pagination.

CRT phototypesetters usually operate in an off-line mode, with computer¬
generated magnetic tape as their input source. The tape contains the same
character, function, and interline spacing codes associated with second
generation devices. Magnetic tape is also used for font loading with those
CRT phototypesetters using digital character storage.

30 Composition Input

The Linotron 505 has an optional capability for magnetic tape input, but is
primarily designed to operate from perforated paper tape input. In many
ways, the Linotron 505 is more closely related to second generation
phototypesetters than to other CRT devices.

Character storage in third generation phototypesetters is either
photographic or digital. Photographic storage requires the positioning of
master character sets in the optical path. Access to fonts not already stored
within the machine requires manual changing.

Digital storage defines character shapes by digital coding read and processed
by a computer. The information required to describe characters digitally is
so large that peripheral magnetic disc systems are required to maintain a
reasonable font selection within control of the computer typesetting system.
Magnetic disc systems are currently used in two ways. (1) directly interfaced
with the main computer system or (2) directly interfaced with the
phototypesetter.

Magnetic disc systems peripheral to the computer enter digital font
information into the phototypesetter by magnetic input tape. Upon entry,
the information is stored in the magnetic core of a small control computer.
Here the information is available on-line as required.

Generation of character images for CRT phototypesetters occurs from
electronic projection of character descriptions stored within the machine.
Phototypesetters that use photgraphic storage employ two CRT’s to
generate character images. One CRT is a scanning device that describes the
character and the second displays the character for exposure of the output
media. Machines that store characters digitally use a single display CRT.

Composition Input 31

PENTA PRISM

POSITIVE FIELD LENS
EYE LENS

OUTPUT MEDIA

REFLECTION

MIRROR

PENTA

PRISM

DECOLLIMATOR

ASSY

OPTICAL
WEDGE ASSY

COLLIMATOR
LENS ASSY

CHARACTER GRID

CONDENSER

ILLUMINATION ASSY

The Linofilm Quick imaging system uses a stationary
matrix.

The Super Quick and Wide Range Super Quick were,
and still are, expanded versions of the basic Quick
photosetter. The unit on the left is the optional Tab-
matic unit for tabular composition.

■■H

;•..

32 Composition Input

Also part of character generation for CRT phototypesetters is the system
used to change point sizes. In scanning systems, height adjustments are
made by changing the length of the stroke. Width changes are more complex
and must be accomplished by either, (1) increasing the dimensions of the
electron beam while spacing the strokes further apart, or (2) maintaining a
constant electron beam size and using more strokes to describe the
character.

CRT phototypesetters designed for photographic character storage vary
point size by using a constant stroke width in the display CRT and
increasing or decreasing the number of strokes used to describe the
character. This is done by varying spacing of the vertical scan lines in the
scanning CRT in relation to the point size. For example, for a 5 point width,
the master character image would be scanned in half as many strokes as used
for a 10 point width. When combined with a constant beam width in the
display CRT, the result is a character width or point size in proportion to the
number of scanning strokes.

Output pagination refers to the method used to generate a full "page of
typeset copy. The primary methods used are: (1) stroke, (2) line, and (3)
page.

In operation, each stroke displayed on the face of the CRT is located in
exactly he same position. No horizontal movement of the electron beam
occurs at any time. A mechanically driven prism lays down each vertical
slice of the character one at a time. A series of timing lines synchronizes
movement of the prism with generation of each stroke.

Composition Input 3 3

INDEX TUBE

The Mergenthaler 1010 imaging system.

The imaging system of the Linotron 505. Note that it
is a hybrid: using a matrix to tell the CRT how a
character should be generated.

The line method utilizes horizontal deflection of the electron beam to typeset
a complete line and a mechanical system to advance the film or paper
between lines. This method is much faster than the stroke method because
electronic beam deflection is faster than mechanical action. However, the
extra speed is achieved at the cost of a more expensive, larger CRT and more
complicated circuitry to compensate for distortion of the electron beam as it
moves away from the center of the tube.

The page method uses both horizontal and vertical deflection of the electron
beam to set a complete area or page of type without any mechanical
movement. The output film or paper is advanced only between pages. While
the fastest typesetting method of all, it is also the most expensive.

The Fototronic CRT uses a combination of the line and page methods. It
combines beam motion with motion of the output film or paper for both
horizontal and vertical placement of characters. Horizontal character
placement is accomplished by beam deflection across the full 11.5-inch width
of the CRT display tube. In addition, the output film or paper is on a
carriage that can be shifted sideways if a wider line is required.

There are several developments in technology and applications that will
significantly affect the phototypesetting industry, although it is too early to
predict the nature or extent of their impact.

34 Composition Input

3. Keyboard to tape systems

To understand the wide variety of keyboards available today, and to
appreciate some of the trends toward new equipment concepts, it is useful to
have some background on the general development of keyboard systems as
used in the typesetting environment. For our purposes here keyboards,
keyboard systems, keyboard machines and keyboard devices are defined as
those pieces of machinery which create a machine readable record as their
primary output. This record is then subsequently processed by some other
device or system. Keyboard elements are those subassemblies consisting of
at least 64 keys mounted on a frame.

Initially, virtually all typesetting was done on hot metal linecasters made by
Harris-Intertype and Merganthaler. The operator used a keyboard
component that was an integral part of the linecaster, and as information
was keyed, a line of brass type mats was assembled to form a line from which
a lead slug was cast. These lead slugs were then locked in a chase and this
formed a complete galley of information ready to be printed. The maximum
line length was limited by the capacity of the linecaster and was usually 30
picas. To achieve uniformity of line, it was necessary for the operator to key
sufficient charaters to fill the line (supply enough characters to form a
complete slug). On the other hand, if too many characters were keyed, the
machine stopped due to an overset condition. This “enough but not too
many” measure (called the hot zone) is simply a method of telling the
operator when enough characters have been keyed to form a line. Once the
hot zone was reached, it was up to the operator to determine how to end the
line-to either finish the word, or to hyphenate.

Three major events moved typesetting away from the use of manual systems
to the use of modern phototypesetting devices of today:

1) The first significant event to influence typesetting procedure was the
demonstration that linecaster could be adapted to operate from punched
paper tape readers. This permitted the tape to be made elsewhere on
relatively inexpensive keyboard systems and then processed by the
linecaster. Three major advantages were:

a) the tape could be prepared by people with less skill than a linecaster
operator

b) the keying speed was faster when only punching tape

c) the linecaster could operate at its maximum speed, hour after hour.

Composition Input 3 5

The use of independent keyboard systems operating a tape driven linecaster
significantly increased the effective “throughput” (or characters per dollar)
of the system.

Several manufacturers offered special keyboard systems to serve this
growing market. The first such devices were blind counting keyboards which
produced justified tape. In use, both the length of the line to be set and the
size of the type to be used was determined and the keyboard device adjusted
for these parameters. As the operator keyed material into tape, an internal
counting mechanism kept track of how much of the line had been consumed.
When the hot zone had been reached, an indicator informed the operator
that it was time to end the line, and the operator had to either finish the word
or hyphenate, and insert a line end code. These blind keyboard systems
consisted of a keyboard element mechanically connected to a paper tape
punch, with a counting mechanism included. The operator had no visibility
of the material being keyed, except to read and translate the code holes
punched in the tape. In an attempt to furnish the operator with visibility of
keyed material, some manufacturers connected tape punches and counting
mechanisms to typewriters, and offered hard copy counting keyboards.
These hard copy systems were more expensive, and in general they found
their best application in training new keyboard operators or in preparing
complex material.

2) The second event concerned the use of computers, which mostly
affected high volume typesetting, such as newspapers. It was found that
operators could key material up to 50 per cent faster if they did not have to
concern themselves with “line end decisions”. By switching off the counting
devices in the keyboards, the operator could key non-justified tape. This
non-justified tape was then fed into a special purpose computer, which had
been programmed to assemble the characters into a specific line length. The
computer also contained a set of logic rules to govern hyphenation. The
computer/then read the non-justified tape, examined each character to
determine how much space in the line it would consume, hyphenated words if
necessary, and produced a justified tape to operate the linecaster.
Compugraphic was the pioneer in the field of computerized typesetting, and
the Justape and Justape Jr. led the field in competition with IBM. The
application of the computer to the typesetting environment permitted an
important increase in keying speed, resulting in an effective lower keying
cost. Manufacturers offered both blind and hard copy keyboard systems to
serve this market application, although in many cases these machines were
simply counting keyboards with the counting mechanisms removed.

These two events greatly increased the effective performance of hot metal
typesetting. The first event (independent tape controlled operation) removed
the human limitation and allowed the linecaster to operate at its maximum
speed; the second event (computerized processing of non-justified tape)
eliminated the concern for line end decisions and allowed the human to key
at appreciably faster rates. For comparison, a good linecaster operator could
key at approximately 6,000 characters per hour, while a good keyboard
operator using a non-counting keyboard could reach rates of 18,000
characters per hour. Thus the use of non-counting keyboards combined with
a typesetting computer achieved a three times increase in typesetting

36 Composition Input

throughout. In addition, the requirement for highly skilled operators was
reduced, which therefore lowered the cost of keyboarding.

3) The third event to consider was a basic change in printing technology and
the resultant impact on typesetting. Hot metal linecasters prepared lines of
type, later assembled into galleys, which were used with letterpress or
sterotype rotary printing equipment.

However, raised type was not usable in an offset press, so an intermediate
step was required between the assembled typeset galley and the offset plate.
Most users simply ran a “repro copy” from the raised type, photographed it,
and used this negative to make the offset plate.

Phototypesetting was born with the development of machines using photo
flash techniques to record the selected character image on a piece of film or
paper. This output film or paper then became the equivalent of the typeset
galley. The character images were selected from font discs or grids through
which the light was flashed and the size of the character image could be
altered by choosing lenses of different sizes, which finally focused the image
on the film. Initial phototypesetters were functional replacements of the hot
metal machines and, therefore, the preparation of input tape was the same
for either hot metal or phototypesetters.

Subsequent phototypesetters offered far more flexibility with respect to font
size and the number of fonts available. This increased the complexity of
keyboarding, and has given rise to more sophisticated keyboard systems to
produce this tape. The role of the computer has also expanded, and in
addition to performing the simple hyphenation-justification tasks, the
computer can now manage many of the phototypesetter functions.

In general, the more straightforward phototypesetters manufactured today
have the hyphenation-justification computer included, and consequently
virtually any kind of tape is acceptable. These machines are mostly used in
straight matter text environments, such as the production of newspapers.
The more sophisticated phototypesetters, which require the use of justified
tape, are usually matched with the complex counting keyboards, or with
computer systems.

Today, a bewildering selection of equipment is available. The user must
choose between counting or non-counting keyboards, a host of typesetting
computers, and typesetters that accept both justified and non-justified tape.
New keyboard systems and computer systems introduce the question of
recording media — magnetic tape, punch tape, printed tape, punch cards —
are all available and they all have their place. As the number of options and
permutations proliferate, it becomes increasingly important to understand
the entire system, and its application. This “system appreciation” is
imperative before an intelligent recommendation can be made relative to the
phototypesetter and its attendant keyboard devices.

Recording Techniques

All recording techniques in use today are based on the absence or presence of
a recording “bit”. A bit, by definition, is a Binary Digit, and binary digits

Composition Input 37

are the fundamental units of a numbering system which uses 2 as a Radix.
This base 2 system (using the “0” and “1” as the digits) is particularly
applicable to recording schemes, since these two digits can be stored or
recorded by a variety of mechanical or electronic devices. For example, relay
contacts can be open or closed, a pulse can be absent or present, a magnetic
field can be polarized north or south, a light can be off or on, etc.

Different numerals, alphabetic characters, or symbols can be recorded by
assembling these bits in various combinations, called codes. The number of
different codes available in any system is a function of the number of levels
used - a level corresponding to a significant place in the system. For
example, a code system using 5 levels has a maximum number of 32 code
combinations. A code system using 6 significant places or levels offers 64
possible combinations, and so on. The number of possible code
combinations is limited only by the number of levels or significant places
used.

Since our concern is with typesetting, our attention will focus on the code
combinations most commonly used by typesetting equipment. The original
code system was developed by the Teletype Corporation, and is referred to
as a TeleTypeSetter, or TTS code. Six levels are available, so the system
offers 64 codes. These codes are punched in paper tape, with a hole being
equal to “1” and a no-hole being equal to “0”.

TTS tape is seven-eighths inches wide, and each of the data holes or bits is
.072 inches in diameter. The sprocket hole is .046 inches in diameter, and
while the primary use of this sprocket hole is to mechanically move the tape
through the punch or reader, it also serves as a reference or timing bit to
identify a character position in the tape. Character density is 10 characters
per inch. Tracks are numbered 0 thru 5 (some manufacturers number the
tracks 1 thru 6) and the sproket hole is in the center of the tape. All the bits
that constitute a character are punched at the same time, along with the
sprocket hole.

As mentioned above, a 6 level TTS code set offers 64 unique code
combinations. Since a typesetter may contain over 100 characters in any
font, plus 10 to 20 control codes, it is necessary to find some way to expand
the number of useful codes available. This is done by using a “precedence”
technique. When a precedence code is punched in the tape, it defines the
meaning of all subsequent codes. The TTS system uses “shift” and “unshift”
as precedence codes. While this does not change the number of unique
combinations in the system, it does increase the number of useful codes,
since any given bit combination can now have two meanings. For example,
in the shift case, a hole in track 0, 1 is the numeric character “ 3 /s”. In the
unshift case the same hole combination is the numeric character “3”. Of the
64 codes in the TTS system, 2 are dedicated as precedence codes, one is the
rub-out code and one is the tape feed code, resulting in 60 codes in the shift
case and 60 in the unshift case, or 120 in all. Some codes, like elevate, return,
thin space, etc., are the same in either the shift or unshift condition.

Some systems use a third precedence code (sometimes called red ribbon shift
or control case shift) which allows 59 codes in each case, or 177 useful codes.

38 Composition Input

Recording Media

Paper tape is one of the original recording media. The paper itself is usually
furnished in 1000 foot rolls, and is normally 2 Vi to 4 mils in thickness. Paper
tape is available in standard widths of eleven-sixteenths inches, seven-eights
inches and 1 inch. While the actual recording format is standard the world
over (data holes .072 inches, sprocket holes .046 inches, holes on one-tenth
inches centers) there are a number of methods used to number the tracks. In
addition, some schemes use the sprocket hole on a line tangent with the date
holes. The scheme used will vary with industry and application, and virtually
any tape width will be found in any industry. In general, the primary
applications of paper tape are:

Feed

Data holes or bits

..07

Sprocket

2 Character Or Row

eleven-sixteenths inches wide
seven-eighths inches wide
seven-eights inches wide
one inch wide

5 level communications

6 level typesetting

7 level data processing

8 level data processing

Paper tape is widely used due to its low cost, coupled with the modest cost of
punches and readers. A 1000 foot roll of tape costs approximately $1.00 and
will contain 120,000 characters. The cost of the punches and readers will
vary with the operating characteristics and quality, with punches of the 15-30
characters per second vaiety costing $350-600 and ranging up to $2500 to
$9000 for punches operating at speeds of 150 to 300 characters per second.
Readers range from $250 to $500 for speeds of 15-30 characters per second.
In general, paper tape finds its most popular application in those areas where
the recording and reading rate is not high, and low cost is an important
factor.

MagneticTape

This concept of recording and retrieving information was pioneered by the
data processing industry, and was designed to satisfy the need for fast input
of data to a computer. The tape is usually furnished in lengths of 2400 feet
(wound on a 10 inch reel) and is Vi inches wide by 1 Vi mils thick. A reel of
tape costs around $35, but the information on the tape can be erased and the
tape reused many times. As with paper tape, the characters are made up of a
pattern of bits. The bits are represented by reversals in magnetic flux
direction. This is done by moving the tape past an electromagnet (called a
write head) and reversing the flow of current at the time a bit is to be
recorded. Heads are arranged across the tape, with one head for each track
or level. Information is recorded by the write head, and is retrieved by the
read head, both of which are contained in the tape deck. Magnetic tape
permits a far higher read-write rate as well as a higher character density than
does paper tape. Today, the most common kinds of tape formats are:

7 track 200, 556, or 800 characters per inch (CPI)

9 track 800 CPI

9 track 1600 CPI (requires special read-write techniques)

Composition Input 39

The tape is moved past the read-write head at a constant speed during the
writing or reading operation. The tape stops in the gaps between blocks. As
the tape is read, a flux change in any of the tracks (or any combination of
tracks) signifies the presence of a character. This makes the tape “self
clocking” and eliminates the need for sprocket bits or timing bits to be
recorded along with the character. The speed at which the characters can be
recorded or retrieved (transfer rate) is a function of the character density
(characters per inch or CPI) and the speed of the tape (inches per second or
IPS). In a system using a tape speed of 40 inches per second, and a character
density of 800 char per inch, the effective transfer rate is 40 IPS X 800 CPI
or 32,000 characters per second, either read or write. Tape drives are
currently available with speeds from 7,000 characters per second up to
180,000 characters per second. Mag tape drives are usually furnished with
both read and write electronics. Depending upon the characteristics of the
unit, a mag tape drive may cost anywhere from $4,000 to $40,000. These
units usually find application in those environments where large amounts of
data must be recorded or retrieved at high speeds, and where prime cost is
not a significant factor, due to large data volume.

New lower cost tape drives (called incremental units) are now being offered
to bridge the gap between the slow rates of paper tape devices and the
extremely fast speeds of the magnetic tape drives discussed in above. These
units record information on magnetic tape in a computer compatible format,
however, in place of recording a block of data at once, they record a single
character at a time. The recording can be done at any random rate up to the
maximum, and the tape increments (moves) one character position for each
character recorded, thus the name incremental. Generally they accept data
at rates up to 300 characters per second, and the tape can then be loaded on a
computer and read at maximum computer speeds. Some incremental
magnetic tape units have both read and write functions, and typically ’’slow-
read” the tape at rates up to 6,000 char per sec. These units sell for around
$5,000 to $10,000, depending upon their characteristics, and are generally
suited for applications where the information is recorded at fairly low speeds
and read into a computer at high speed.

Another approach to the problem of low cost recording coupled with fairly
fast reading speeds is based on the use of !4 inch magnetic tape. This tape is
similar to that used in home entertainment systems, with an important
difference being the characteristic quality of the oxide coating. As before,
characters are made up of groups of bits, and (like Vi inch mag tape) the bits
are recorded by changing the flux direction as the tape moves past the read

GaP D □□□□□□ □ Gap □ □ □ Etc - 1/4"

Bits 12345678 123 i

A representation of seven track magnetic tape.

40 Composition Input

head. Bit patterns are recorded sequentially in tracks. The tape pattern
provides a start-stop gap between characters, so that the system may record
and read information a character at a time. Generally these systems are all
based on an 8 level code, thus 8 bits are always in the string. The system is
self clocking, so no sprocket or timing bits are recorded. The recording
density can be as high as 800 bits per inch, but since the characters are
sparated by gaps, the effective density is around 40 characters per inch.
These units offer recording rates as high as 1600 char per sec. The price of a
combined read-write unit is approximately $3,000.

Magnetic cassettes also evolved out of the home entertainment field,
and were pioneered by Philips-Norelco. They are extremely easy to handle,
and are very small in size considering the amount of information they can
store. In principle, they are identical to the !4 inch mag tape machines in
that they record the bits in serial strings, with gaps between characters.
Cassettes use tape only Vs inches wide. They can store up to 80,000
characters in a single cassette, and the tape is re-usable many times. The cost
of a single cassette varies from $3 - $12. Cassettes are widely touted as the
replacement devices for punched paper tape, and there are examples where
the cost of the recording and reading units appear competitive with punched
tape. A unit combining both read and write functions (such as the Sykes
unit) may be priced as low as $750, and offers write speeds of up 500 char per
sec. Their primary disadvantage seems to be the lack of an industry wide
standard governing the record-format. Various manufacturers of cassette
equipment may use different track-bit-code allocations, which means that
cassettes may not be interchangable between devices made by different
companies.

While the term cassette (by popular definition) means the Phillips-Norellco
style, more and more companies are offering their own unique versions.
IBM uses a special cassette on their MT-ST-MT-SC equipment, where the
tape is Vs inches wide and is moved by means of sprocket holes punched in
the edge of the tape,similar to motion picture film. Characters are recorded
in parallel across the tape, similar to punched paper tape. Invac offers a
special cassette using l A inch magnetic tape, where the tape moves from reel
to reel, and the characters are recorded in bit sequential fashion.

Keyboard Primer

This section deals with keyboard systems that are now in common usage.
Some of these keyboards have been available for many years, while others
are relatively new. It is impossible to compare every feature of every
keyboard to its competitor, so this section is intended to give an overview of
keyboards currently in use.

Keyboard element — For hard copy devices, this means the kind of
typewriter used. For blind devices, this refers to the keyboard component.

Secretary shift — This defines the shift and unshift configurations of the
keyboard element. Some manufacturers use a separate key for both shift and
unshift commands with the attendant code generated as either key is
depressed. Secretary shift commonly means that the shift code is generated
when the key is depressed, the unshift code when the key is released.

Composition Input 41

TAPE IDENTIFICATION CHART

PUNCHING

DIES

CHANNEL

NUMBERS

CHANNEL —

FRAME**-

ADVANCED _

FEED HOLE

RUBOUTS <

TAPE FEED <

ADD 7 —
ADD 8 —

TAPE PUNCH

-►OOOOO o ooo

128

)

ONE INCH TAPE

FLOW
OF TAPE

FEED HOLE .4375"
FROM GUIDE EDGE

GUIDE EDGE

BINARY VALUES

*A BIT IS EITHER THE PRESENCE OR ABSENCE OF A HOLE
**A CODE MAY BE ONE OR MORE FRAMES

SW>C£ H 8f>A«£ 1 SPACC

~fis m mvt m -My

zma § ucao ■ te*»:
ueao 1 ipr 1 mm

fan at

plash 1 mat # wi
onu 1 just I; mm

ill

: WHiai r

■ ' i i

: TA

•

n§

- m

III 4-i

4 k

Datek of England manufactures an extremely ex¬
tensive line of input keyboards and peripheral equip¬
ment. One peripheral that is of increasing interest is
the line printer. This is the Datek version of a device
to accept input tape and print out its contents for
proofreading purposes.

Additional keys on the Dual Image keyboard permit
single stroke command of all Compugraphic 4961TL
typographic functions. Depression of one key often
produces two printed codes.

mmvx

we* cm,c;

imz izmm

m m

*: c a :

cm# um.

A closeup of the key set of the Mergenthaler V-I-P
keyboard. Note the extra rows of character keys. The
indicator in the center displays line length for
justification purposes. This is counting keyboard.

Composition Input 43

Back space — This feature allows the tape to be moved back one character
at a time from the keyboard element. Most keyboard systems have some
provision for reversing the tape, but not all do this by depressing a key.

Repeat key — This key, when depressed with certain other keys (or in some
cases, any other key) causes that particular key to repeat until released.

Format storage — This is the ability to store, either on tape or
electronically, a number of predetermined character sequences for changing
line length, leading, font size, etc.

Multiple codes - one key — This is the ability to generate a 2 or 3 character
sequence from a single key on the keyboard element. These are usually pre¬
wired at the factory and relate to some specific typesetter function, for
example ‘flash only’.

Widths mixed from the keyboard — Some keyboards can store different
character width values (usually in the form of width plugs) and the operator
can select the one desired by merely pressing a key. Width plugs usually
relate to a particular font.

Code format — This refers to the number of channels or holes in
punchepaper tape. Where an entry appears it means that the unit is available
in either 6 level TTS code, 7 level Friden, or 8 level ASCII.

Hot zone indicator — The method used to tell the operator when the line end
decision should be made.

Maximum line length picas — While true line measure varies with the
counting technique used, most counting keyboards are classified by the
maximum line length that can be counted, as measured in picas.

Type of display — In some cases, a typewriter; in others a moving display of
the last 16 characters.

Number of characters — Refers to the number of different symbols that can
be displayed to the operator, and has nothing to do with the number of
characters in any given font set.

Notes on counting

To ensure that the finished typeset galley is well balanced and asthetically
pleasing, it is necessary to use a system of differential character widths. In
this way the larger characters, such as the W and M, occupy more horizontal
space than the smaller characters, such as the I and J. Two methods are in
use today.

Unit cut fonts

Sometimes called standard, or standard cut mats. In this system, the widest
character in the font (the upper case ‘M’, also called the ‘em quad’) is divided
into 18 equal parts, and multiples of the one eighteenth dimension become
the width of the different characters in the font.

44 Composition Input

Non-unit cut fonts

One version of the Datek tape perforating keyboard
(marketed in the United States by VGC). The
keyboard is noiseless and the punch is covered.
Operators were raising the cover in order to hear
some indication that a key registered. Manufacturers
have found that this “sound of accomplishment” is
quite important.

Sometimes called multiface mats, this system is based on the division of the
em quad into 32 units. Multiples of the one-thirty second dimension become
the width of the different characters in the font. For example, the lower case
T may be 7 units (seven-thirty seconds of the em) with the upper case ‘M’
being a full 32 units. This ratio of character widths is true only for a given
font size. That is, the T may be 7 units in 12 point type, but only 5 units in a
16 point font. The multiface system allows greater precision in typesetting,
since the width relationship of the characters varies with the point size of the
font.

Blind, counting keyboards

Fairchild was the first company to offer keyboard systems to prepare the
tape. The keyboards were designed to match a given linecaster, and were
tailored to the kind of font being used, (unit cut, or multiface). Where the
keyboard was used with a unit cut standard font, the keyboard was simply
adjusted for the length of the line and the size of the type. Where the keboard
was used with a multiface font, a ‘width plug’ was inserted to program the
keyboard for the character width relationships, and the keyboard was
adjusted for the length of the line to be set. Each font had an appropriate
width plug. These machines were intended for the production environment
and very little flexibility was built into them. With multiface fonts, it was
necessary to manually change width plugs when changing font sizes.

Subsequent keyboards offered more flexibility, and multiface machines were
now offered with the capability of having more than one width plug inserted
at once. The Fairchild Universal 210 could hold two, and the Merganthaler
Lino-Quick could hold 4. This permitted widths to be mixed from the
keyboard, since the desired width plug could be selected by simply pressing a
key. Blind keyboards offer no visibility of the data being keyed, although
some operators are proficient at reading the holes in the tape. Most
keyboards furnish indicator lights to display the status of the shift condition,
upper or lower rail, etc. Line end decisions are made by the operator once
the hot zone is reached. Most of these keyboards can be used in the non¬
counting mode.

Notes on hand copy

There are a number of different viewpoints on the subject of hard copy, and
the opinions are varied and somewhat subjective. In essence, hard copy
offers the keyboard operator visibility of the material being keyed. Most
hard copy devices use a typewriter, which means that a typed page is
prepared which corresponds to the information being recorded. Typewriters
can display upper case, lower case, numbers and symbols. The typewriter
character set is limited to the number of character positions available. The
IBM Selectric machine furnishes 88 positions, while some of the type bar
machines offer as many as 96. Since the printing is done by one mechanism,
and the recording by another, it is impossible to guarantee that the
information being typed is exactly the same as the information being
recorded. Because of the character set limitations it is not feasible for a
typewriter to display control codes, such as upper rail, quad left, etc.

Composition Input 45

Proponents of the hard copy machines feel that the primary advantage is
accuracy. Since the operator can see every character, it is very easy to
correct the most common keying problem, the single keystroke error. The
hard copy machines are very useful for training new operators, and in
addition, it is easier to compose complex copy (like display advertising)
when the information is visible. The opponents of hard copy maintain that
visibility of the data actually slows down the operator, and faster sustained
keying rates are possible if the operator can not see what is being keyed. In
addition, the necessity of returning the carriage breaks the operator rhythm
and reduces keying rates. (In counting keyboards, where the operator must
observe a hot zone indicator and make line end decisions, this argument is
hard to support. It’s extremely valid with non-counting keyboards).

The trade-off seems to be between higher production speeds (blind
keyboards) and increased accuracy (hard copy systems). The final
configuration is determined by the kind of task.

Hard copy, counting keyboards

These devices were originally offered to furnish more accurate keying of
tapes to drive hot metal machines. However, the advent of phototypesetters
began to change the keyboarding problem, due to the greater flexibility of
the phototypesetting systems. It was now possible to set both straight matter
and ad copy on phototypesetters, however the keying task to accomplish this
became much more complex. For example, it required only a single
keystroke (Upper Rail) to go from normal face to bold face, since the line
length, leading, and character width criteria remained the same. However, to
go from 8 point to 16 point type could take as many as 20 keystrokes in a
precise sequence, because of the line measure implication. Thus, keyboards
were offered to address the problems of doing in-line width mixing, and the
changing of character fonts and-or line length. These keyboards contain
subroutines that are adjustable to match a number of different typesetting or
composing tasks.

The same counting procedures are used by these devices as with the blind
counting keboards. Line end decisions are made by the operator once the hot
zone is reached, and lights are used to display the status of the shift, upper or
lower rails, etc. Most units can be used to produce non-counted tape.

Blind, non-counting keyboards

The advent of the typesetting computer created the market for these
keyboards. The primary intention was for the keying of straight matter,
where the operator made no line end decisions and speed was of primary
importance. These keyboards permit the fastest posssible keying rate, and
are limited only by the speed of the operator. While there is no visibility of
the material being keyed, experienced operators seldom find this a problem,
although new operators do. These are the simplest keyboard systems
available, consequently they are the lowest priced. Some of these keyboards
can be connected to slave typewriters to furnish hard copy, but this is not
common in a production environment. Status lights are furnished to display
the shift condition, upper or lower rails, etc.

46 Composition Input

A second type of blind non-counting keyboard was offered to prepare tapes
for the more sophisticated phototypesetters. Rather than build format
storage routines into the keyboards, the complex routines were stored in the
computer used with the phototypesetter. In this way, four or five codes
punched in the non-justified tape could cause the computer to insert the
proper sequence of commands in the justified output tape. Keyboards with
this feature usually have the function keys on a separate auxiliary keyboard
and are sometimes referred to as ‘computer’ keyboards.

Hard copy, non-counting keyboards

These keyboards are generally counting keyboards with the counting
mechanisms removed, or else typewriter systems that were originally
intended for other reasons and are now being offered to the Graphic Arts
Industry. They cost more than blind devices, and do not permit the same
high keying speeds due to the necessity of periodically returning the carriage
which breaks the operator ryhthm. Their primary application is the more
accurate keying of straight matter, or the training of new keyboard
operators.

Once the material has been typeset and reviewed by the proof reader and the
appropriate areas of the action identified, the information is passed on to the
place where the material is corrected.

Irrespective of the nature of the error, it will always be corrected by one of
two actions: (a) something must be inserted, and-or (b) something must be
removed. With these two correction steps in mind, we now need to address
the reason for the change.

This involves the correction of keystroke errors, or spelling mistakes. To
implement the correction, it is necessary to have access only to that portion
of the material in question. There is no need to be able to review either
preceding or subsequent material. Further, the material can be corrected by
the substitution of very small amounts of new data; usually the insertion of a
correct letter or correct word is all that is required.

This involves the complete overview of all the material, or at least the
overview of some complete segment of it. Editing may require entire
paragraphs removed, replaced, or relocated elesewhere in the sequence. To
implement revisions, it is necessary to have access to large amounts of
material, and the amounts of new data may be considerable.

This applies to such problems as updating telephone directories, personnel
rosters, guidebooks, etc. The material in question is typically handled in a
unit (e.g. name and address, name and department number-telephone, etc.)
and whole units of information are removed or inserted. Access to large
amounts of the material is not required, since the information is usually
arranged in sequence (alphabetic or numeric) and the insertions or deletions
can be structured the same way.

The factors that influence corrections are: amount of material to which
access is necessary, size of insertions-deletions (number of characters or

Composition Input 47

words), method used to insert material, time permitted to complete
alternation.

The process of correcting, editing, and-or updating material is inherent in
any typesetting operation. It is useful to understand the traditional
procedures before examing some of the newer devices being offered. Since
proof reading is virtually the same in all cases, our primary focus will be on
the manner in which corrections and-or alterations are handled.

Manual Methods

Proofreading is done from the galley. The galley may be the output paper
from a phototypesetter, or it may be a reprocopy pulled from the chase
containing the cast lead slugs. After proof reading, the fix can be
accomplished in two ways:

Hot Metal

The correct information is keyed, and new slugs or lines are cast. These slugs
are inserted in the chase in place of the incorrect or non-valid lines. The
minimum amount of material that can be replaced is one full line.

Phototypesetters

The correct information is keyed, and the tape run through the
phototypesetting machine. The output paper is then pasted over the incorrect
or non-valid portions of the original galley. In practice, all the corrections
are keyed and processed at the same time, so the output paper may contain a
number of corrections.

If the correction involves the insertion of some information previously
omitted, it may be necessary to rekey entire paragraphs of correct
information because of the ‘domino effect’ of the insertion.

With either method, it is impossible to proof read until after the typesetting
function. In the event of a number of errors, the usual procedure is to re-key
all the information and discard the first pass. The lack of visibility of the
data prior to typesetting often results in a significant amount of waste, in
terms of people time, machine time, and supplies cost.

Automated Methods

A number of systems are available today, offered in different configurations
and with varying degrees of flexibility. Basically these machines offer the
same principal thing; the facility to display and massage the material prior
to typesetting. The methods used to display and the flexibility of the
alteration procedure are the primary considerations when evaluating the use
of these systems. Some machines permit proof reading and correction to be
done at the same time, other systems are structured so that correction takes
place on a second pass. Some systems can address several different types of
alterations.

48 Composition Input

It is very difficult to make meaningful comparisons of different types of
systems, due to the varying capabilities of the hardware and the extremely
broad range of correction - editing - updating problems. Rather than make
comparisons, it seems more reasonable to illustrate the principle of a given
system as applied to a given alteration task. The best system for a given
application depends on the application and the economics governing it.

Keyboard methodology

The earliest tape perforating keyboards were completely mechanical. On a
mechanical keyboard the keys are linked to the tape punch by a series of
levers. Under each key, and running from the front to the rear of the
machine are coding bars which are serrated in accordance with the code to
be punched, and these operate, a series of combination bars which run at
right angles to them. There is a combination bar for each code channel, and
they interpose a system underneath the punch knives. When a key is
depressed, the required code is set on he interposes under the punch knives
and at the same time a clutch is operated which causes a continuously
running electric motor to operate a cam under the punch knives. This cam
forces the knives through the paper tape in those channels where the
interposes have been set on the code bar.

There is necessarily a ball-lock device on the mechanical keyboard to
prevent two keys from being depressed, simultaneously. This can also be
adjusted to operate early or late on the keystroke. With the continuously
running motor, the mechanical movements make the keyboard noisy in
operation. Because of the type of construction involved, the slope of the key
panel of this type of keyboard is usually much greater than the 12 or 15
degrees which has been found to be optimum on modern high-speed typing
devices. Also, the space between keys is generally 7/8”, as against the 3/4”
spacing of typewriters. Because of the mechanical design it is impossible to
change the layout of the keyboard or to add additional keys, so that these
machines are very inflexible.

In an effort to overcome the problems of slow and heavy operation of the
completely mechanical keyboard, electromechanical keyboards were
introduced. These keyboards still operate coding bars from the keys, but the
combination bars of the mechanical keyboard are eliminated. The operation
of a key allows the code bar to slide forward and operate on a series of
contacts. These contacts are wired to the code magnets of the punch which
set up interposers under the punch knives. The key also operates a clutch
which allows the continuously running electric motor to turn a cam and force
the punch knives through the paper. The mechanical interlock has been
taken off the actual key and operates on a sliding code bar on some types of
these machines so that only one bar is allowed to make contact at one time.

The third type of machine is the contact type. It employs a simple electrical
contact or switch which, when the key lever is depressed, brings two pieces of
electrically conductive material into contact, allowing a current to pass. For
the contact to break when a key is released, it is usual for a contact to be
mounted on very spring-responsive material. Because of this, whenever a
contact is made by depressing the key, ‘contact-balance’ occurs. That is, the

Composition Input 49

two contacts make and break contact rapidly. This causes two problems:
first, balancing-contacts can allow a code to pass after the key has been
released, and this can lead to a corrupted code being punched when the
following key is operated quickly. Second, whenever a current is broken by
means of a contact or switch, arcing takes place. To a lesser extent, arcing
also occurs when the contact is made. In time, the arcing builds up a high-
resistance deposit on the contacts and thus reduces the current flying
through the contacts.

To overcome the known problems of switching current, a photo-electric
keyboard was produced by the Invac Corporation. This type of keyboard
consists of a number of light channels with a photo diode at one end and a
light source at the other. There is a light channel for each bit of the code plus
a control channel, and they run parallel to the back of the machine. Over the
light channels and at right angles to them are a series of code bars which are
operated when a key is depressed. These code bars are designed in such a
way that a shutter drops into the light channel when the key is operated, thus
preventing the light from reaching the photo diode at the other end. The
number and position of the shutters on the code bar determine the code to be
punched by that key, since the photo diode only passes current when it is
illuminated. The count is used to actuate the punch knife in that particular
bottom of its stroke, the shutter would not reach the bottom of the light
channel. This would allow light to pass through and cause a wrongly
punched code. To meet this intrinsic problem in the photo-electric
keyboards, the keys are power-assisted so that when a key is depressed a
solenoid or electromagnet is operated, it pulls down a key plus its code bar,
thus insuring that the shutters reach the bottom of the light channel.
Unfortunately the force with which the shutters hit the bottom of the light
channels causes them to vibrate, thus allowing spurious light onto the photo
diode. This spurious light can, of course, cause mispunching when the
keyboard is operated quickly.

In an effort to overcome this, a very heavy mechanical interlock had to be
introduced, and its slows the keyboard down considerably. Consequently, it
can be seen that ‘contact-balance’ has been replaced by ‘shutter-flutter’ with
the same damaging results. Another cause of trouble on the photo-electric
keyboards is the gradual reduction in the strength of the light source on the
output from the photo diodes. This can eventually lead to mispunching. As
with mechanical and electro-mechanical machines, there is no flexibility of
keyboard layout, and if additional keys are required, they have to be built
into a supplementary keyboard.

Finally, there is the reed switch type of keyboard. It incorporates keys
consisting of a relay operated by a magnet. When the key is depressed, the
magnet is moved down the reed and causes the contacts of the reed relay to
close and so allow current to pass. When the original keyboard of this type
was introduced some years ago in Europe it was thought that here at last
might be the answer to the problems on mechanical, electro-mcchanical,
contact and photo-electric keyboards. However, the initial installation was a
failure because the operators complained emphatically about the feel of the
keyboard. An operator likes to know that when he or she presses a key that it
has operated effectively, and this can only be verified by a feel that
something has happened at some some point in depressing the key. On a reed

50 Composition Input

Dual Image Keyboard

Dual Image keyboard devices must be considered in a separate category, due
to the difficulty in comparing them to any traditional keyboards. Dual
Image offers the speed advantages of blind noncounting keyboards,
combined with all the visibility benefits of hard copy machines.

The system utilizes electronic keyboard elements which are solid state
devices. With the exception of the key shaft itself, they have no moving
parts. A two character memory is provided so that burst keying does not
result in any lost or erroneous characters, and/ this is coupled with a
recording mechanism that can accept data at speeds up to 30 characters per
second. The keyboard, the memory, and the recorder combine to form a
keyboard system that can operate at keying rates in excess of 100,000
keystrokes per hour.

The main keyboard of the Dual Image Recorder.

The use of printed paper tape gives all the benefits of hard copy, yet does not
introduce any of the disadvantages of typewriters. All other hard copy
keyboard systems depend upon the use of a typewriter for visibility,which
has two major disadvantages: 1) limited character set, 2) need for carriage
return, which breaks keying rhythm and reduces effective keying speeds.

FInally,.pass.It.on.to.your.machines.at.high.speed.

** * ,, •• ji •j 1 i.. i «•*. B • ■•* • , . i.f

Dual Image tape.

Composition Input 51

- i

'llllSlll:

Characters and functions are immediately visible
when keyed on the Dual Image keyboard. Dual Image
is an interesting approach to the hard copy concept.

Dual Image is not affected by either drawbacks, since the machine can
display a full set of 128 different symbols. The printing is on a continuous
strip of tape, so the operator never needs to break keying rhythm.

It can display both upper and lower case textual material, just like a
typewriter. In addition, the expanded character set permits control codes to
also be printed, and these control codes describe the function so that the
operator needs to make no 'meaning 1 translation. For example, Quad Left is
QL, Upper Rail is UR, Lower Magazine is LM. Other hard copy machines,
using typewriters, frequently print nothing for the command codes.
Sometimes a command code is a textual character printed in red.

52 Composition Input

The heart of the Dual Image system is the high speed impact printer. The
tape to be printed is positioned between a hammer and a high speed print
wheel (which rotates at 1800 rpm). When the desired character is in position,
the hammer strikes, printing both the human readable symbol and the
machine readable code at once. There can never be any disagreement
between what the human reads and what the machine reads. The last
character printed is immediately visible to the operator, so that material
being keyed can be immediately reviewed and confirmed. The recorder
operates with a minimum of moving parts, ensuring a long life and trouble
free operation.

Once the tape is prepared, it can be easily read into a computer or
phototypesetter by using a Dual Image reader. This is a solid state photo-
optical scanner, and can read information at rates up to 300 characters per
second. As the tape is moved past the lens, the bit pattern is focused on a
plane of photo-transistors, and the character is electronically read. The
reader can be made to emulate a variety of traditional devices, such as the
Teletype CX 100 reader, the Tally 424, or Digitronics readers.

CONTROL

AREA

Fig. 1 Proposed U.S.A. Standard: Logical Bit Pairing

CONTROL

AREA

the typewriter key set as used for computer input. The
program was sponsored by the American National
Standards Institute (ANSI) and the Business Equip¬
ment Manufacturers Association (BEMA).

Composition Input 53

How to tailor a keyboard. Here is one company that
makes the job fairly simple. Invac provides a layout
grid for indication of keytops and codes.

Combination Chart Used for All
INVAC Keyboard Formats

INVAC Corporation provides three basic
keyboard formats:

(1) PK-244 (48 keys)

(2) PK-264 (64 keys)

(3) PK-275 (75 keys)

Three shades of gray indicate the limits
of each keyboard format (see Legend
below). Note that the PK-275 extends
across all three shaded areas , the PK-264
uses the two lighter shades, while only the
central, lighter area comprises the PK-244.

Each key position is connected by a line
to a row of code and key color specification
spaces in the lower section of the chart.
Please note that individual "key-numbers"
identify both the key position and the
related code and key color data.

Please refer to the notes on this page
and the "Check List" on Page 1.

A sample chart on Page 4 shows a typi¬
cal customer specification.

Other Information Needed to Complete
the Specification

A step-by-step check list has been provided
on page 1 to facilitate specifying the key¬
board format and code data. Space has
been provided on page 4 for additional
remarks.

Notes:

1. Three keyboard formats are available: 48
keys, 64 keys, and 75 keys. In addition,
certain keys may be arranged in either
diagonal rows (as shown) or vertical rows
(such as one finds on an adding machine).
Unless otherwise specified the diagonal
keyboard will be provided. See Table 1,
page 1 for special key size. Please enter
special key size in Check List under
“Special Features”.

2. Use key numbers for any special refer¬
ences such as key sizes, colors, special
contacts, etc.

3. For convenience, and to minimize error,
please designate code bits by repeating
bit number in the appropriate space. See
illustrated example on page 4.

4. See Table 2, page 1 for available keytop
colors. Unless otherwise specified. Dark
Blue (“B") keytops will be provided. Please
enter this data in the appropriate spaces
provided on the Format Chart.

5. See Table 3, page 1 for available keytop
fill colors. Unless otherwise specified,
White ("W") keytop fill will be provided.
Please enter this data in the appropriate
spaces provided on the Format Chart.

FORMAT CHART for INVAC Keyboards, Codes, and Keys.

LEGEND FOR KEYBOARD FORMATS

mu +

n +

= PK-275 Keyboard

jj +

= PK-264 Keyboard

= PK-244 Keyboard

54 Composition Input

Code and Key Data (See notes designated by superscript numbers)

4. Input media and coding

The Greeks were first again. About 300B.C., Polybius reported the following
system: stations were erected at many locations which consisted of two walls
about seven feet long and six feet high, separated by a space of three feet. At
night, one or more torches, as needed, but no more than five, were placed on
top of the walls. Certain combinations of torches represented Greek letters.
Two torches on the right wall and three on the left may have stood for the
letter 4 H’ as an example. Thus, words were spelled out letter by letter on this
five-unit, center-feed “tele”-torch communication system.

In 1887 Herman Hollerith constructed the first electromechanical system for
recording, computing and tabulating digital data, which he then used to
record the 1890 census. Holes were punched in cards with a conductor’s
punch. These cards were positioned over a series of mercury-filled cups and
at the touch of a lever, telescoping pins projected to the card’s surface and
then through, if there was a hole. The pin reaching the mercury completed an
electrical circuit and and this in turn moved a pointer one position on a dial.
The punched (or punch) card was one of the first methods of recording
information.

A punched card measures 7 3/8 by 3 1/4 inches and is 0.007 inches thick. It
contains 80 columns (the 90 column card developed by Univac was
discontinued in 1966) which are numbered from left to right, 1 to 80.
Vertically there are twelve rows numbered 12, 11, and then 0 to 9, from top
to bottom. The two top rows are also called the X(11) and Y(12). The upper
left-hand corner may be cut, or the corners rounded, depending on the
system utilized. An alpha or numeric character is represented by holes
punched in one or more locations of a single row. Groups of characters, or
rows, such as columns 1 through 30 or 23 through 34 for example, are called
fields. A field is a unique group of characters and may represent a part
number or a description or an address. Thus all addresses may be put in
certain columns, and we always know (as does the computer) where to find
the address for comparison, sorting or correction purposes.

The punched card idea goes back to the early 1800’s. Punched cardboard
patterns were used to direct textile looms by mechanically selecting hooks
which raised the longitudinal threads to make a passage for the shuttle,
which “set” the crossthreads. J.M. Jacquard thus automated the weaving
process and paved the way for the automatic tape operation of industrial

Composition Input 5 5

machinery that would come 150 years later. The punched hole was also used
by the Monotype as a tiny “valve” which controlled air pressure and in turn
positioned a matrix case.

Punched cards are an almost permanent recording medium. They may be
retained for long periods and thus eliminate re-punching. Each card is a
“slice” of information, a record, and may be changed without affecting other
cards. The punched card began the unit record concept.

The early telegraph, from 1850 to 1920, used the Morse code of dots and
dashes. The duration of a dot was one unit; that of a dash was three units.
There were also three units between characters and six units between words.
J.M.E. Baudot applied the idea of two early telegraphic developers, Gauss
and Weber, of a code using five units to represent the alphabet, numerals and
special symbols. This system, with some variations in character allocations,
became the standard telegraphic code system. A seven unit code was devised
by J.B. Moore which permitted detection of code mutilation by transmission
and reception devices. A modified form of the seven unit code was later
expanded by ELC.A. VanDuuren.

In 1858 Sir Charles Wheatstone employed a perforated tape to operate the
telegraph transmitting mechanism. At first it was driven by a clock device
and later by an electric motor. The Morse signals were received by an
“inker” in which an inked wheel marked the dots and dashes on a moving
tape.

' 3

Em Ld|

CL.

Cl.

(Si

j=.

>■

’S)

CtL

□

•

■

□

■

□

■

□

■

□

■

□

■

□

•

□

■

□

■

□

•

□

■

•

□

■

□

■

□

■

□

■

□

■

□

■

□

L°

•

□

■

□

■

□

■

□

■

□

■

r -

■

□

■

□

■

□

■

□

■

□

■

□

■

□

■

□

■

□

■

□

■

□

■

□

■

□

■

□

■

□

■

□

■

□

■

□

■

□

■

□

■

□

The TeleTypeSetter code.

In 1915 Western Union established the multiplex system of printing
telegraphy. It provided a number of independent channels of communication
going in both directions at the same time on a single wire. Here, punched
tape was used extensively for the first time. The Baudot five unit alphabet
was used primarily. Five impulses of negative and positive current were
combined to form a given letter. Thus A was represented as two positive and
three negative units.

Punched tape may be paper or plastic. In either case there is always one long
lengthwise row of small sprocket feed holes and five, six, seven or eight rows
of larger parallel holes which represent character and function codes. Here
are some of the types and forms of punched paper tape:

Oiled: tape impregnated lightly and uniformly with oil for lubrication and
ease of punching.

Dry. paper tape, period.

Mylar, plastic tape which is more durable and may be re-run more often
Strip: two to four foot sections of tape.

56 Composition Input

Fan-folded: tape folded every six or so inches into a stack.

Roll: about 700 feet of tape in its most used format.

Center-feed: the sprocket holes are lined up with the middle of the code
holes.

Advance-feed', the sprocket holes line up with the leading edge of the code
holes. This approach was developed to avoid confusion with which end goes
first. When the sprocket hole is to the left of the code hole (on a right to left
reader) the tape is being read correctly.

Magnetic tape is polyester plastic with a coating of magnetic particles.
Invented by O. Smith in 1880, who impregnated a cotton thread with steel
dust, mag tape took fifty years to perfect. Today’s tape is about one half inch
wide, one half mil to one and a half mils thick and 2400 feet long. Quarter,
three quarter and even inch tape in various lengths is also used.

Information is recorded on the tape by magnetizing narrow strips called
tracks in alternating directions. Thus, one frame of paper tape with holes
representing binary codes is represented on mag tape by the presence of a
magnetic flux reversal; no holes are represented by the absence of a flux
reversal. 200,556, or 800 bits per inch (bpi) or code frames may be recorded.
The width and bpi of mag tape is determined by the tape reader employed.
Tracks are presently designated as either seven-track or nine-track, and each
apply only to certain computers. A mag tape cassette is a length of magnetic
tape wound so as to form a continuous loop, with an opening at which the
tape may be read and recorded.

Each hole in a card or tape or flux reversal on mag represents a bit. Bit
stands for binary digit and means yes or l or on if it is there, and no or zero
or off if it is not. Here’s where a little arithmetic comes in. We use a
numbering system every day that is based on the number 10. In the number
567 the “7” position is digits (total: 7); the "6” position is tens (6 x 10 equals
60); the “5 position is hundreds (5 x 100 equals 500); so the number is
expressed as 567. Thus we count by multiplying the number of tens in the
position in which a number occurs. Going from right to left:

10 6

10 5

10 4

10 3

10 2

1,000,000

100,000

10,000

1,000

100

V-I-P

CODE

V-I-P

FUNCTION

5 4 3 2 1 0

• • • •

EM LEADER

• •*•••

EN LEADER

• •

SPACEBAND

• • • •

EM SPACE

• • • • •

EN SPACE

•

THIN SPACE

••••••

UNSHIFT

• • • ••

SHIFT

• • • •

SUPERSHIFT

• • • ••

QUAD LEFT

QUAD RIGHT

••••• •

QUAD CENTER

• • • •••

LOWER RAIL

• • • ••

UPPER RAIL

• • •••

BELL

• •

ELEVATE

• • •

ADD THIN

• •

RETURN

• • •

PAPER FEED

•••••••

RUBOUT

•

TAPE FEED

Command codes for the Mergenthaler V-I-P.

10 1
10

(10 0 )
( 1 )

Composition Input 5 7

The above review is more than you need to know about binary arithmetic (so
why tell me after I read it?); but it is a necessary introduction to coding. To
encode only the numerals 0 through 9 would require codes four bits long:
0000= 0, 0001 = 1,0010= 2, 0011 = 3, 0100= 4, 0101= 5, 0110= 6, 0111 =
7, 1000= 8, 1001 = 9. Actually sixteen combinations are possible; not nearly
enough to encode an entire alphabet. A five bit code, such as the teletype
code, permits 64 possible combinations. Still inadequate for typesetting. The
teletypesetter utilizes a six bit code with 128 possible combinations.
Advanced typesetters require even more codes than this.

TAPE

CHANNEL

NUMBERS

CHARACTER

FUNCTION

Ell

UNSHIFT

SHIFT

■

fli

Eli

wrm

mom

•

■

mom

mm\

hhi

Oil

Ofll

Efl

Dll

EMI

mam

EMI

VMI

mrm

■HI

mfli

Efl

♦

Dll

EMI

log

•

gg !

ggpj

«bH

worn

MTi

Efl

gOM

■

wrm

g g

Efl

IDfl

■

Efl

•

—

■

•

mrm

•

i&ig

mrm

worn

•

Efl

•

■H

♦

wrwwt

ROH

♦

ggg

wrm

mrm

•

man

mrm

!■

'/.

•

HHI

mrm

♦

WWM

wrm

mom

♦

arm

mrm

mom

3 A

mom

«r

■fli

IHflHI

mom

♦

IHSEEHl

arm

mom

- -'—

wrm

mrm

mom

worn

BBS

mrm

worn

flflfl

COMMA

flggQQZflHI

mrm

•

wrm

mrm

;

■EX

mom

•

mom

mm.

mmm

mum

)

(

worn

APOS.

QUOTE

mom

■OH

■HR

'KM

HYPHEN

mtm

ggg

r ~

mmm

ggg

TAPE

FEED

—

arm

mom

ROM

EN SPACE

mom

flOP

mrm

igH

arm

EN LEADER

mom

flOP

♦

mom

iggg

flH

EM SPACE

■mi

•

mom

iggg

EM LEADER

BOR

:ggg|

\mm

mmm

Iggg

IKM

VERT. RULE

mom

1 |ggg

IMM

IRHRg

THIN

SPACE |

ggg

imm

Iggg

SPACE BAR

ggg

;PSgi|*

IKM

IRHHH

RETURN

mom

I ggg

IRHfl

ELEVATE

mrm

wrm

imm

iggg

IflH

PAPER FEED

Efl

■ihhi

IKM

I ROH

SHIFT

Efl

ggg

IKfl

IKM

in!

UNSHIFT

arm

IHMC

IKfl

IROHI

UPPER RAIL

mom

mrm

Efl

IMM

IKfl

IflH

1 LOWER RAIL

mom

Kfl

flH

IH1M

IKM

IRHfl

STOP

r *i

mom

ggg

I MTM

IKfl

iwnm

RUB

OUT |

ggg

Iflfll

IRHfl

ADD THIN

mmm

ggg

IKfl

Imm

QUAD LEFT

mom

flH

Kfl

IKfl

IKS

} QUAD

CENTER j

ggjI

mom

IKfl

IHH

QUAD RIGHT

•

ggg

yaggggg

f mmm

Iflfll

IRRRg

* ADD 7

•

Iflfll

iflHH

ADD 8

Coding chart for the Friden 8201 keyboard.

58 Composition Input

Because the majority of phototypesetters standardized on the TTS 6-level
code due to the proliferation of keyboards, newer devices require multiple
codes to access certain characters and functions. Thus, some keyboards
produce six-level tape with a seventh or eighth bit key to make no code more
than two keystrokes. Others produce 8-level tape which provides a more
complete code array.

Coding chart for the Friden LCC-VF keyboard.

TAPE

CHANNEL

NUMBERS

CHARACTER OR FUNCTION

FEED

UNSHIFT

Lower Rail

Upper Rail

Lower Rail

Upper Rail

•

ffi

•

ffl

•

Period

•

Comma

•

Semicolon

Colon

•

)

(

•

Apos./Quote

Quote

•

—

•

TAPE

FEED

•

EN SP.

EN LD.

•

EN LD.

•

EM SP.

EM LD.

•

EM LD.

EM SP.

•

VERT. RULE

•

SHIFT

•

UNSHIFT

•

UPPER RAIL

•

LOWER RAIL

•

STOP CODE

•

CODE DELETE

•

r QUAD LEFT

•

K QUAD CENTER

•

QUAD RIGHT

•

THIN SPACE

•

SPACE BAR

•

CARRIAGE RETURN

•

ELEV. ELEV.

•

PAPER FEED

•

■-

•

SPACE BAR and THIN SPACE

HOT METAL EQUIVALENT

Composition Input 59

VALUE UNSHIFT 5 4 3 2 10 SHIFT SUPER SHIFT

0 0

"IF

■n

0 1

■

□

HiH

III

■

□

3/8

•

■

e? msm

ADD POINT

•

□

•

ran

•

mmm h

»&m:-nnii

■

□

■

□

[KILL LINE

1 0

■

1 I

□

1 2

■

□

■

□

EM-

1 4

■

1 5

□

Sgsii

1 6

RETURN

□

■

1 7

APO S »

□

1 8

1 9

HY-

2 0

■

□

2 1

'/2

2 2

t§

■

•

2 3

COMMAND

2 4

□

WTW

COMMA

□

1 1 1 11

VY1

□

Q L

□

a l

2 8

■

□

■

□

3 1

Q R

□

Q ft

•

mamm

KOI

r 9

□

mum

III

•

□

■

□

*/.

□

EDI

SUPER SHIFT

•

■

□

SUPERSHIFT

III

□

fDI

□

'/4

EM LDR

□

•

EM LOR

WWW

□

•

□

i U

wrm

□

•

0 ZERO

•

o tua

•

EN LDR

•1

•

□

EN LOR

•

■

□

•

□

•

□

U R

•

□

•

□

;

SHIFT

□

•

□

L R

•

•1

•

•i

PERIOD

■

□

•

PERIOD

• l

•

■

□

•

□

'/s

■

•

WMM

Q C

•

Q C

UNSHIFT

•

•1

•

UNSHIFT

RUBOUT

■

DEC

VALUE UNSHIFT 5 4 3 2 10 SHIFT SUPER SHIFT

”4"

•

IKSH

•

ELEVATE

•

ELEVATE

cc ec

ZUPER SHIFT

•

! A

UJ UJ

CIRCUMFLEX HTCl

•

CIRCUN

SPACE

IFLEXju^

NOTE (UC) ACCENT FOR UPPER CASE CHARAC

(LC) ACCENT FOR LOWER CASE CHARAC

0 8

0 9

SPACE BAND

BAND

“DOT LESS 1

1 DOTLESS I

i i

_ TT

□

1 2

•

■

1 3

FIG -

□

1 4

□

<•

1 5

1 6

■mill —

1 7

APOS 1

□

1 8

■

1 9

HY-

•

■

□

*/•

2 0

■

□

111

■ 111111 rn^rn

□

■

Wl EH

COMMA

□

■

•

Q L

■

•1

•

Q L

•

□

MlIB

3 0

■

•

3 1

1 UNIT SPACE

■

□

1 UNIT SPACE

■

□

■

□

•

□

■

□

■

•

□

' GRAVE (UC)

1 3 8

mmnwmwmm

■

□

Lai

DIERISIS (LC)

•

EMFIP3]

WWW

□

WWW

□

■

□

0 ZERO

_s_

□

h i im hi ii

□

'ACUTE(UC)

■

□

■

C 1

5 1

■

5 2

□

•

□

•

SHIFT

•J

•

SHIFT

L R

□

•

■

□

1 58

• □

■

□

•1

•

[6 1

Q C

•1

■

□

rwm

UNSHIFT

RUBOUT

□

i _

■

STD. TYPESETTING CODING

TAPE

CHANNEL

NUMBERS

CHARACTER

FUNCTION

□

UNSHIFT

SHIFT

•

HDH

•

MMBMMM

•

H j

•

o !

SB!

•

HHEHEH

•

HBEBBB

•

HQBl

•

H 1

•

■■■■■E

•

COMMA

•

)

I I

•

APOS.

QUOTE

•

HYPHEN

-t-

TAPE

FEED j

•

EN SPACE

Iks

•

EN LEADER

•

EM SPACE

•

EM LEADER

•

VERT.

RULE J

•

THIN SPACE

•

SPACE BAR

CAR RETURN

•

ELEVATE

•

PAPER

FEED

•

SHIFT

(UC)

•

UNSHIFT (LC) !

•

UPPER

RAIL |

•

LOWER RAIL

•

STOP (BELL) CODE

•

■1

RUB

OUT

mrm

•

ADD

THIN

•

QUAD

LEFT |

■lb

•

CENTER 1

•

QUAD

RIGHT

WMM

ADD 7

ADO •

Standard typesetting coding according to Friden. The
“6” channel becomes the “0” channel in TTS.

Coding for the V-I-P keyboard comes in two versions:
left, the U.S. and British, and right, the European ver¬
sion.

60 Composition Input

In this section you will find some coding systems used on present day
devices. Note that most are based on the teletypesetter 6-level code with
modifications. In any case, the coding structure controls the number of
keystrokes needed to access any character or command.

Since machines make mistakes (people are perfect) an additional bit is
provided in some systems as a doublecheck on proper functioning of the code
translating device, whether reader or writer. This is the “parity” bit. Using
one and zero as the basic building blocks of any code produces code chains
with an even or uneven number of ones. A parity bit is added to make the
“byte” (a byte is a complete set of bits forming one word) even or odd (as the
system dictates) and this then allows the system to detect errors and thus
assure accuracy.

Character coding for the Mergenthaler V-I-P.

Mirror image coding is used for certain input systems;
it “flops” the tape. Thus the character ‘a’ which was
54 is now 32.

MIRROR IMAGE CODING

TAPE

CHANNEL

NUMBERS

CHARACTER

FUNCTION

FEED

UNSHIFT

SHIFT

•

J /«

•

’/.

•

Em dash

•

COMMA

•

)

(

•

APOS.

QUOTE

•

HYPHEN

TAPE

FEED

•

EN SPACE

•

EN LEADER

•

• -

•

EM SPACE

•

EM LEADER

•

VERT. RULE

•

THIN SPACE

•

SPACE BAR

CAR RETURN

•

ELEVATE

•

SHIFT (UC)

•

UNSHIFT (LC)

•

UPPER RAIL

•

LOWER RAIL

•

STOP (BELL) CODE

•

RUB OUT

•

•-

•

ADD THIN

•

QUAD LEFT

•

QUAD CENTER

•

ADD 7

ADD 8

• ••

• • • •
• • • •

• • •

• •
• • • •
• • • •

• • •

• • ••
• • • ••
• • •

• •• •

• ••

• • •

• • • •

• • • ••

• • •

• • ••

• •• • •

• ••

• •• • •

• • • •

• • •
•••• ••
• • •••
• ••
• • • •
► • •
• • • ••
• • •••
• • • •
• • • •
• • • •
• •• • •
• • • •
• • • • •
• • •

» • ••
• • ••
• •••

All

B12

A13

D12

C15

D14

D13

A12

B16

Cl 2

B11

B14

A14

D15

A16

C11

C14

B13

Dll

DIO

A15

C16

B15

D16

D17

B17

D21

Cl 7

B21

V-I-P STANDARD FONTS

UNSHIFT

u. o
o

C13

CIO

LAYOUT

P4S-1 P45-2 P45-3 P34-1 P34-2

CHARACTER

PERIOD

COMMA

HYPHEN

PHOTON
INSTRUCTIONS

LENS POS 1

LENS POS. 2

LENS POS. 3

LENS POS. 4

LENS POS. 5

LENS POS 6

LENS POS 7

LENS POS. 8

LOWER PAIR

LINE LENGTH

SHIFT

o <
u- O
o

A10

BIO

A18

B18

C18

D18

A19

B19

C19

D19

A20

B20

C20

A17

C21

A21

D20

LAYOUT

P4S-1

P45-2

P45-3

P34-1

P34-:

CHARACTER

y«

Em Minus Em Em I Em

+ H

PERIOD

COMMA

% + % / /

SUPERSHIFT

A22

B22

C22

D22

A23

B23

C23

D23

A24

B24

C24

(A21)

D24

LAYOUT

P45-1 P45-2 P45-3 P34-1 P34

CHARACTER

* « <

% + %

* * *

CODE +
7 BIT

THIN SPACE

Min

□

☆

® |TM

Composition Input 61

Character and command coding for the Mergenthaler
Super Quick.

Two newer code systems are being used with greater frequency in an attempt
to standardize information transfer. The first is the United States of
America Standard Code for Information Interchange (USASCII or ASCII)
and the other is the Extended Binary Coded Decimal Interchange Code
(EBCIDIC). Each provides a larger array of code possibilities than are
possible on TTS, the unofficial standard of the industry. These new code
systems provide a unique code for each character or command rather than
shift, unshift precedence coding (two frames for one capital letter, for
instance).

62 Composition Input

5. Punched card input systems

The RACE II System from Warlock Computer Corporation uses punched
cards rather than perforated paper tape as the means of communication
between the typesetting keyboard and the phototypesetter. It does not limit
the capabilities of the keyboard but adds the functions of a data processing
system, such as:

Random Access — The random access capability makes it possible to locate
any text line or group of text lines by simply reading the legend at the top of
the punched card. Once located, text lines may be changed, deleted, or
rearranged as desired.

Editing and Updating — The ability to easily edit or update material such as
price lists, directories, parts lists, etc., which are subject to periodic reruns
with changes.

Manipulation — The ability to manipulate material to be typeset using high
speed card sorting (unit record) equipment where the size or economics of
the job do not justify fully computerized storage and manipulation. In this
way, a variety of reports to be typeset may be produced from one carefully
organized deck of cards.

Computer Access — The ability to typeset material which is already stored
and manipulated in a standard business computer directly without requiring
special hardware which is not readily available.

The RACE II Card Input System consists basically of two components, the
Keyboard Interpeter Unit (KIU) and the Typesetter Interface Unit (TIU).

Keyboard Interpreter Unit — The Keyboard Interpeter Unit (KIU)
provides the means for generating punched cards, using a standard
typesetting keyboard. The unit serves as a link between a typesetting
keyboard and an IBM Keypunch, Model 026 or 029.

Typesetter Interface Unit (TIU) — The Typesetter Interface Unit is used to
transmit data from punched cards to the Phototypesetter system.

The KIU is connected directly to the punch drive circuit of the typesetting
keyboard. As the operator of the keyboard prepares the copy, data in the

Composition Input 63

code format of the keyboard is routed through the connecting cable to the
KIU. The KIU circuitry converts the data from the code format of the
keyboard to Hollerith coding and then drives an IBM Keypunch to enter the
data on cards.

The Console of the TIU is interfaced between the tape reader of the
phototypesetter and the logic circuits of the typesetter. Controls on the
Console enable the operator to select either paper tape or cards as the input
medium to the phototypesetter. When in the paper tape mode of operation,
the signals from the tape reader are routed directly through the output cable
to the phototypesetter. When in the card mode of operation, signals from the
card reader are routed to the Converter Cabinet of the TIU where the
Hollerith formatted signals are converted to TTS code or appropriate
typesetter code format. The signals are routed through the Console to the
phototypesetter which reads the data as though it originated on paper tape.
The Console also makes it possible for the operator to select the column on
the card where reading will begin, the column where reading will end, and
the first and last columns within the card which identify a block of
information to be deleted.

A standard eighty column punched card.

Character or row

Information

Sections

An IBM System 3 punched card.

Tracks or Leve

Each keystroke at the keyboard will cause one column to be punched on the
card. The alphabet and the numerals are printed at the top of the card with
alphabetic characters appearing as capital letters regardless of the shift or
unshift condition of the keyboard at the time. In addition to the alphabet and
numerals, the period, comma, hyphen, fraction bar and several special
characters print out at the top of the card. The special characters are used to
indicate the most frequently used function codes. This makes it easier to
determine that a card has been prepared correctly than trying to read the
punched codes. Some function codes will be punched on the card but will not
print a symbol at the top of the column.

64 Composition Input

The function codes for line length, leading, point size and typeface are
keyboarded as though the information were being entered on tape. These
instructions should be followed by a carriage return to insure that the line
parameters are on a single card which may be replaced, if in error, or
duplicated, with the duplicate cards inserted in the deck where needed to
avoid keyboarding the information each time it is required. Normal straight
matter is keyboarded just as if the information were being placed on paper
tape, terminating each line with a carriage return. This will automatically
release the card and register a new one ready for the next line. Under certain
circumstances, it is possible for the operator to get ahead of the keypunch
after a carriage return, resulting in the loss of codes on the card. This occurs
when the carriage on the keyboard completes its return to the left margin
faster than the keypunch can release one card and register a new one. The
operator should develop the habit of coordinating the typing operation with
the keypunch rather than the keyboard. Lines exceeding eighty codes must
be continued on another card. If the line can be terminated before the end of
the card, no futher attention is required. If the line must be continued on the
next card, it is necessary for the operator to depress the REL key of the
keypunch, wait until the new card has registered and then complete the line,
ending with a carriage return.

Certain tabular runs may have columns to be deleted in some galleys and
added in others. For example: (a) A large number of items that require
automatic card sorting in accordance with coded categoried i.e. brass, iron,
plastic, wood parts, etc. (b) Wholesale/retail prices to be produced from the
same deck, (c) Customer billing information such as in classified
advertisements. In this case, the Card Length and Field Delete switches are
used to identify the fields (columns) on the card which are to be typeset. If
the information to be deleted is at the front of the card, set the Card Length
Begin switch to identify the last column to be ignored. If the information to
be deleted is at the end of the card, set the Card Length End switches to
identify the last card column to be read and the Field Delete End switches to
identify the last column to be deleted. The system will run faster if
information to be deleted is at the end of the card rather than the beginning.
This is due to the fact that the phototypesetter begins flashing a line
following carriage return while the reader ejects the card, inserts a new card,
and then waits for the “START” signal from the phototypesetter before
reading the new card. It is more efficient to allorbefore reading the new card.
It is more efficient to allow the card reader to pass over deleted information
while the phototypesetter is flashing, than to force the phototypesetter to
wait during this period.

In the normal mode of operation, keystrokes are duplicated on cards column
by column. That is, each keystroke produces a corresponding punched code
in the card. The alphabet and the numerals are printed at the top of the card
and since the keypunch has only one “case”, the legend at the top of the card
will always appear as capital letters or numbers regardless of the “shift” or
“unshift” condition of the keyboard.

Composition Input 65

In addition to the alphabetical characters and numerals, periods, commas,
hyphens, fraction bars and several additional special characters are also
printed on the top of the card. The special characters are used to indicate the
most frequently used function codes which makes it easier to determine if a
card has been correctly punched.

It is a matter of convenience whether or not function codes are placed on the
same card with the text matter or are placed on separate cards. Placing the
function codes on the same cards with the text will sometimes reduce the size
of a deck significantly as well as eliminate the possibility of misplacing of
these codes. On the other hand, if function codes are placed on separate
cards, it will be easier to change formatting of instructions such as leading
and font, in case this should be required after initial keyboarding is
completed i.e. increasing or decreasing the leading point in order to achieve a
better copy fit.

The flexibility of the keypunch and the RACE II Card Input System as well
as the variety of keyboards which can be used with this system make it
impossible to anticipate all possible combinations of tab requirements. The
following paragraphs, therefore, describe a few typical examples in order to
illustrate the basic principles of tabular work.

Examine the list for longest part number, longest nomenclature, or longest
retail and wholesale prices, including the dollar sign. Keyboard the
composite “longest line” using fixed spaces between part number and
nomenclature, and between retail and wholesale prices, and use J-spaces
between nomenclature and retail price. A test setting of the text line can be
made to determine the final line length to be used in the typesetter or else,
the line length can be calculated from a character count and/or the character
width tables available in the appropriate typesetter Manual.

From this card, a program card for the keypunch is made up which causes
the RACE II Card Input System to perform repetitive operations
automatically.

The sequence of subsequent operations is as follows:

(1) The keypunch registers a card, punches the lower case code and then
stops.

(2) The operator keyboards the part number and depresses the SKIP key.

(3) The keypunch punches the tab codes, shift code and “Quad Left” code
for the nomenclature column, and then stops.

(4) The operator keyboards the nomenclature, shifting and unshifting as
necessary, and depresses the SKIP key.

(5) The keypunch skips over the remaining columns of the nomenclature
field, and then punches in the tab codes and a “Quad Right” code for the
retail price column and stops.

66 Composition Input

(6) The operator keyboards the dollar sign and/or fixed spaces plus the
retail price, as required, and depresses the SKIP key.

(7) The keypunch punches in the tab codes, the “Quad Right” code for the
wholesale column, and stops.

(8) The operator keyboards the wholesale price in the same manner as the
retail price, described in (6) above, and depresses the SKIP key.

(9) The keypunch releases the card, registers a new card, punches the unshift
code, and stops, ready for the next line.

HOLLERITH CODE

TTS CODE

CHARACTER

HOLLERITH CODE

CHARACTER

TTS CODE

11 12

•

—

•

—

•

—

•

—

•

—

•

—

•

—

•

—

•

—

•

—

•

—

•

—

•

—

•

—

•

—

•

—

•

—

•

—i

•

—

•

—

•

—

•

—

•

—

■■

•

—

•

—

•

—

•

—

•

—

—i

•

—

•

—

•

I—

—

•

—

•

—

•

—

•

—

•

—

■■

•

A conversion chart of Hollerith (punched card codes)
and TTS coding.

□

□1

[

CARR. RETURN

•

■

—

•

□

■

—i

PERIOD

•

■1

□1

—

LINE LENGTH

•

■1

□1

—

■

■I

EM SPACE

•

□

—

□i

■1

■

—

EN SPACE

•

□

□1

—

QUOTE

•

—

•

■1

■

□

■

ADD LEAD

•

—

•

□

—

•

□

—

(

•

■

□

■

—

COMMA

•

■

□

■

—

BELL

•

—

■

□

■

BLR

•

—

■

-»

EM LDR

•

■

□

■

EN LDR

•

—

HYPHEN

•

—

■

UPPER CASE

•

—

■

LOWER CASE

•

□

■

UPPER ROW

•

. LOWER ROW

•

□

—

QUAD LEFT

•

—

■

QUAD CENTER

•

□

■

—

QUAD RIGHT

•

—

SEMI COLON

•

—

THIN SPACE

•

■

—

IU SPACE

•

—

■

—

EOL LEAD

•

—

SPACE BAND

•

■

LENS ONE

•

—i

■

LENS TWO

•

■

—

LENS THREE

•

■

—

LENS FOUR

•

■

—

■

LENS FIVE

•

■

LENS SIX

•

—

■

LENS SEVEN

•

LENS EIGHT

•

■

TAPE FEED

•

—

TAB (713-5)

•

■

—

TAB (713-10)

•

■

STOP

•

Composition Input 67

100111 BRAHMS - Sonata In F Minor For Clarinet And Piano/$13.
Sonata In E-Flat Major For Clarinet and Piano -
Harold Wright, Clarinet; Harris Goldsmith, Piano

The fixed information in this example is much greater than the variable
information. Furthermore, for the purposes of ths example, it is also
assumed that the variable information changes frequently. Examine the copy
for the longest part number and the longest price. Either a test setting can be
made, or else the width can be calculated from a character count and/or
character width table. The space remaining can be devoted to the
nomenclature and description columns. Since the price changes frequently
and only represents a small part of the total text, it is best to place it on a
card by itself. Keyboarding would proceed as follows:

(1) With the regular program card on the keypunch and the starwheels
down, keyboard part number, “TAB”, the first line of text matter to
justification, “TAB” and “Quad Right” codes, and depress the “REL” key.

(2) Keypunch registers a new card.

(3) Operator keyboards the price and a “Carriage Return”.

(4) Keypunch registers a new card.

(5) Operator keyboards “TAB”, the second line of text matter to
justification, and a “Carriage Return”.

(6) Keypunch registers a new card. Additional lines of text matter are
keyboarded as in (5) above until the text is completed.

(7) Start the second item as in (1) above.

Price revisions can be keyboarded most efficiently in one continuous
operation and be inserted into the card deck in a second operation. This will
reduce greatly the possibility of human error. The use of colored cards can
be a help in locating price cards or in keeping track of new versus old prices.

The following is an explanation of the terms used in the RACE II Card
Input System.

Column

The RACE II Card Input System uses standard 80 column cards. A column
has twelve possible punch positions, sometimes referred to as “twelve
levels”. Each column is equivalent to one code frame on perforated paper
tapes. Columns are numbered 1 through 80 with column numbers normally

68 Composition Input

printed on the face of the card. In conversation be careful to distinguish
between a “card column” and a “column of tabular copy”.

DUP

This stands for “Duplicate”. The use of this keypunch key permits
duplication of information from one card to another. However, it does NOT
duplicate information from the program card.

Field

A group of columns on a punched card may be referred to as a “Field.” For
example, a group of columns on a punched card which contains a price might
be referred to as the “Price Field”. Similarly there may be a “Date Field”, a
“Name Field”, etc.

Format Card

This term refers to a card which contains format instructions for the
typesetter. Do not confuse it with a Program Card used on the Keypunch.

Line Card

A card which contains nothing but a carriage return (line space) code is
called a “line card”. It produces an amount of leading which is equal to the
primary leading programmed into the typesetter.

Program Card

A program card is used on the program drum of the keypunch to control the
automatic functions of the keypunch. No information is transferred from the
program card to the card currently being punched. The program card
excercises control over which columns of information are to be duplicated
from the card in the read station to the card in the punch station, or which
columns are to be skipped entirely.

Program Drum

This is the drum in the keypunch which holds the program card. When the
starwheels are down, the program card controls the automatic functions of
the keypunch. When the starwheels are in the up position, control of the
keypunch is performed manually.

Rel

This key on the keypunch causes the card in the punch station to be released
and a new card to be registered.

Row

The cards used by the RACE II Card Input System have twelve “rows”. The
“twelve” row is located at the top of the card, immediately below the printed

Composition Input 69

legend. The next row then is
“four”, “five”, “six”, “seven”,

“eleven”, “zero”, “one”, “two”, “three”,
“eight”, “nine”, in that order down the card.

Run-Out Card

A card with sufficient “add-leads” to advance the paper in the magazine
after a run on the typesetter is used in place of the manual leading.

Skip

Depressing this key on the keypunch causes the card to advance rapidly to
the next field as defined by the program card. With the program card used
for normal text, the card will skip the remainder of the card, and a new card
will be registered in the punch station.

Stop Card

This is a card with a “stop” code plus a “tape feed” (blank frame or column)
code which is placed at the end of a deck of cards to be run. If it is at the end
of a take, it should be preceded by a “run-out” card.

The Mergenthaler V-I-P, which stands for Variable
Input Phototypesetter. It incorporates its own
minicomputer.

70 Composition Input

6. Video display terminal (VDT) systems

There are over 200 video display terminals on the market. Some are the
alphanumeric type and show numbers, letters and special symbols on the
screen. Others are “graphic” terminals with the ability to draw lines on the
screen and thereby display representations of three-dimensional objects.

The video display terminals used in typesetting today are alphanumeric,
varying from simple models which can display only a few hundred cap
characters to those which may indicate different “typefaces” on the screen.

A video display terminal is a special kind of television. An electron gun
projects a stream of electrons onto a phosphor coating on the face of the
tube, causing the tube to glow. The color of the image varies with the kind of
coating inside or the tinted screen which is over the front of the tube.

The phosphor coating is excited to a state of fluorescence that has a degree
of persistence (it takes a time for the image to fade). If the phosphor fades
quickly the image fades likewise. Because the image projected on the face of
the tube begins to fade it must be refreshed or displayed over and over again
for the image to appear stable on the face of the tube. This means that the
information to be displayed must be held in some sort of buffer memory and
recycled many times a second to achieve a reasonably flicker free image.

Video terminals may be configured in a wide variety of different ways. There
is the totally self-contained editing terminal which does not need a computer.
This unit reads text in from paper tape, displays it on the screen to be
modified and then repunches the corrected version as a clean tape.

There is also the slave unit with virtually no logic of its own. This unit must
be wired directly to a computer. In between these units is the stand-alone
computer terminal connected to a computer to accept or deliver hunks of
text, but with its own logic for storage of the text to be displayed on the
screen and for making changes in this text. Lastly, there is the cluster
terminal which shares logic among several separate stations.

Characters are formed on the face of a VDT tube in the following ways:

1. Characters are created out of a series of dots. Most are designed to form a
rectangle five dots wide by seven dots high (a 5 x 7 dot matrix). It is difficult
to design legible lowercase letters within these constraints and some VDT’s

Composition Input 71

have gone to a 7 x 9 or greater matrix. A dot matrix pattern is well suited to
a television tube with its fixed raster scan. This can, in effect, make any TV a
terminal.

2. Characters are created out of a series of line segments. This is called
vector generated characters and is less commonly used.

3. Another approach is to “paint” the characters with a series of horizontal
or vertical strokes (raster scan), in a manner similar to that used on most
CRT typesetters — although, of course, with much lower resolution.

The number of characters which may be displayed on the tube at one time is
a function of the resolution of the tube. The more addressable points there
are on a tube, and the fewer addressable points required to describe each
character, the more characters will fit on the screen at one time. The size of
the tube itself is of relatively little consequence in determining how much
information will fit on the screen. You get the same picture information on a
7” T.V. tube as you do on a 21” tube. You may prefer the 21” tube because
it is larger, but all you are really seeing is an enlarged and more legible
version of the same picture.

Most of the VDTs which are commercially available are not aimed
specifically at the typesetting market and have limited character sets —
often about 64 separate symbols, which means upper case characters only
and very few special symbols. Increasingly, greater numbers of upper and
lower case alphanumeric terminals are coming on the market. Most of these
will display over 96 separate symbols. Some will display as many as 128.

There are three basic elements common to VDT units: a) Refresh logic, b)
Character generation, and c) Editing electronics. As characters enter the
VDT via keyboard, paper tape or by direct cable, they are stored in the
refresh logic. From here the characters are projected on the face of the
cathode ray tube sixty times a second. This cycle is necessary so that the
image on the screen appears stable. The shape and characteristics of each
character are produced by character generation circuitry.

On most units, the character location on the screen at which the on-going
operation takes place is defined by a “cursor.” This may be an underline
dash or an entire rectangle, often having a “blinking” appearance for ease in
location on a screen full of characters. The cursor is the key to most VDT
editing. It is positioned by means of control keys that move it up, down, left
and right. Some VDTs also have a “home” key to bring the cursor to the
first location in the left hand corner of the screen. Additonal commands may
move it to the end of a paragraph or to other locations. It is also used to
delineate the character, word, line or paragraph under editing scrutiny.

VDTs incorporate standard typewriter (QWERTY) sets of keys with two
extra sets: one for editing and one for typesetting functions.

Here is a list of the editing and function controls and what they do for one
specific VDT. This list is by no means inclusive, and presented only to orient
you to the kinds of command keys available on VDTs.

72 Composition Input

CONTROL or
INDICA TOR

FUNCTION

SUPPRESS

Switch

Toggle switch. In up (Suppress) position inhibits display
of function symbols on screen, but does not delete them
from memory. In down (Normal) position, allows func¬
tion symbols to be displayed.

JUST/UNJUST

Switch

Toggle switch. In JUST position, causes Elevate code
read from tape or keyboard to move cursor to new line.
In UNJUST position, Elevate codes are displayed and
punched, but are not executed.

EDIT/PERF

Switch

Toggle switch. In EDIT position, all functions are ac¬
tive. In PERF position, characters typed on the keyboard
are punched one by one. The screen and tape reader are
disabled.

BRIGHTNESS

Potentiometer

Controls brightness of characters displayed on screen.
Rotate clockwise to increase brightness, counterclock¬
wise to decrease brightness.

TAPE LOW

Lamp

Lamps lights when tape-out switch on punch tension
plate opens, indicating that only a few inches of blank
tape remain. Punch is disabled, but unpunched text
remains intact in punch buffer. By setting EDIT/PERF
switch to PERF, leader codes may be punched before
loading new spool of tape.

K/BSHIFT

Lamp

Lights whenever machine enters shifted mode; goes out
when machine returns to unshifted mode.

RESET

Switch

Pushbutton switch. When pressed, erases entire screen,
drives cursor to Home position, sets unshifted mode,
and initializes reader and punch interface.

QL Key

Displays Quad Left symbol and drives cursor to start
of new line.

QC Key

Displays Quad Center symbol and drives cursor to start
of new line.

QR Key

Displays Quad Right symbol and drives cursor to start
of new line.

UR Key

Displays Upper Rail symbol.

Composition Input 7 3

RUBOUT Key Displays Rubout symbol.

LR Key

Displays Lower Rail symbol.

RET Key

Displays Return symbol.

BELL Key

Displays two separate symbols for shifted and unshifted
modes.

Alternate action key; if machine is in unshifted mode,
SHIFT LOCK SHIFT LOCK sets shifted mode and lights K/B SHIFT
Key lamp; if machine is in shifted mode, SHIFT LOCK sets

unshifted mode and extinguishes K/B SHIFT lamp.

ELEV Key Displays Elevate symbol. In JUST mode only, drives

cursor to start of new line.

TAPE FEED Key Displays Tape Feed symbol; has no immediate effect

on punch.

HOME

Drives cursor to Home position (top left corner of
(screen).

(Cursor Up)

Key

Drives cursor up one line for each pressure. Inhibited if
cursor is already in top line.

DEE START

Key

When pressed, current cursor position is defined as start
of punch or delete block operation. Cursor must then
be moved to indicate end of operation.

(Cursor Left)

Key

Drives cursor one character position left for each pres¬
sure. Inhibited if cursor is already in leftmost position
of line or column.

NEW LINE Key

Drives cursor to start of next line.

(Cursor Right) Drives cursor one character position right for each pres-
Key sure. Inhibited when cursor reaches end of screen.

(Cursor Down) Drives cursor down one line for each pressure. Inhibited
Key when cursor reaches bottom line.

Starts tape reader. Reader stops automatically after
READ TAPE approximately 1600 characters have been read and dis-
Key played at 50 cps. rate. After the halt, reading may be

continued by pressing READ TAPE key for each char¬
acter or by holding key down to read at Repeat rate of
15 cps.

Halts automatic reading at any time. In automatic read
STOP READ area of screen, may be held down while READ TAPE
Key key is pressed to read single characters.

-^—i---

74 Composition Input

PUNCH TAPE
KEY

When pressed, a block of text is transferred to the punch
buffer and punched. The start of the block is defined by
the cursor position at the time the DEF START key is
pressed; if no start position is defined, punching starts
from the Home position. The end of the block is the
character to the left of the cursor position at the time
PUNCH TAPE is pressed.

INS CHAR

Key

The character at the cursor position and all characters
to the right of it are moved one position right; a null is
inserted at the cursor position. The function is inhibited
if the shift would move any character except a space into
the last position of the line.

DEL CHAR

Key

The character at the cursor position and all characters
to the right of the cursor are moved one position left.
A null is inserted in the last position of the line.

OPEN Key

All characters from the cursor position to the end of
text are recopied starting at the end of screen and build¬
ing up.

CLOSE Key

Material at the bottom of the screen is recopied, starting
at the cursor position. This closes any gap left after in¬
sertion of new material; the wraparound feature prevents
the breaking of words at line endings.

DEL BLOCK
Key

Erases all characters between a previously defined Start
(or the Home position if no Start was defined) and the
current cursor position. The gap is closed.

CLR Key

Erases all characters from the current cursor position
to end of screen. Nulls are inserted in the erased
positions.

A full view of the CorRecTerm keyboard. There are
three basic, if undemarcated areas: typewriter charac¬
ter set, typesetter function set and display control set.

Composition Input 75

Editing

When a character or group of characters is presented on the screen, certain
editing functions are possible, more or less dependent upon the particular
VDT. Most errors involve single characters, and these are corrected by
positioning the cursor at the character position involved and depressing the
correct character key. This erases the incorrect character and replaces it.
The technique is called “writing over.” On some units it may be necessary to
first strike a DELETE key to erase the incorrect character, hit an INSERT
key and then the correct character. Writing over is much simpler.

Characters are either inserted, deleted or changed. So are lines and
paragraphs. Thus editing keys may be provided to perform each of these
functions. In all these operations the cursor is used to define the data to be
operated on. Positioned at the beginning of the word or paragraph, new data
may be inserted or removed. When deleted, most units automatically close
up the resultant space by moving all characters to the right of the “hole” left
to fill it up. Insertion of characters moves all data right to make room. An
important function of VDTs is the ability to “wraparound”; that is, take
words that will not fit on a line and move them to the beginning of the next
line and so on until the end is reached.

The operator has now input all data, modified it through changes, additions
and deletions and is now ready to be output in the form of tape or direct
impulses to a photo unit.

Stand-alone units

Take one VDT “tube,” put a reader on one side and a punch on the otherand
you have a stand-alone unit. You can read in tapes produced on other
keyboards, or type in information from the keyboard. The resulting data
may be reviewed and edited and then output via the punch to produce a new
tape.

Expanded systems

If a great deal of data will be edited, a magnetic disc or drum or tape cassette
may be used to store all characters before and after their screen debut.
Normally a “controller” which is often a mini-computer directs character
traffic between the storage medium and the VDT.

Expanded VDT systems may also involve dependence upon larger general-
purpose computers. Here much greater capability is available for massive
amounts of editing, storage and retrieval.

VDTs began as what data processing folks call “data windows.” The first
units were replacements, not for keypunches or typewriters, but for
teletypes. Teletypes are widely used for computer input. Systems progressed
and were tied into computers for reservations, credit checking and inventory
control. The late, great company called Viatron attempted to create a unit
that would pre-process data prior to computer input. The year Viatron bit
the dust over forty-five VDTs were exhibited at the Spring Joint Computer
Conference. Many more were certainly in the wings.

76 Composition Input

It was a little under three years ago when the first VDTs were shown to
anxiously - waiting newspapers at the ANPA/RI exhibit in Chicago (1969).
It has taken some time but we are just getting started.

The Hendrix 5700 video display.

Some important points:

The size of the screen is important. Both for the amount of data to be
handled at one gulp and to the clarity and size of characters. VDTs are a
visual medium, and like television, can contribute to eye fatigue. One of the
larger units today has an 18 inch (diagonal) screen that displays 30 lines of
90 characters. About 2,000 characters is the average.

Erasing the screen gets rid of material which is no longer needed. Used
carelessly, it will also get rid of material which is needed. Character, Cursor-
to-end-of-line, and Line Delete functions handle short erasures.

Scrolling may vary from simple movement of displayed material upward, off
the top of the screen both upward and downward movement.

Scrolling depends largely upon the relationship between the VDT’s refresh
logic and its screen complement. When memory capacity equals the number
of characters which can be displayed, memory content is exactly the same as
the display. Blanks are stored in memory for all unused positions on the
screen. More sophisticated units are designed so that only the characters
which are displayed take up space in the refresh memory.

Composition Input 77

Here is the keyboard arrangement of one VDT system presently on the
market. Note the extensive editing capabilities available.

78 Composition Input

s c ur Sort Keys C&*?4r)

1. CLEAR MEMORY (MAIN)
COPY MAIN INTO AUXILIARY

25. ADD THIN SPACE
EM SPACE

2. COPY AUXILIARY INTO MAIN
SWAP MAIN AND AUXILIARY

26. QUAD RIGHT
EN SPACE

3. LEFT MARGIN CONTROL 27. RETURN CURSOR (entry)

4. RIGHT MARGIN CONTROL 28. CURSOR RETURN (editing)

5. DISPLAY MODE SELECTOR

6. RESET SYSTEM LOGIC

29. CHANGE CASE

INSERT MODE SELECTOR

30. CHANGE BOLD

7. INTERLACE ON/OFF BOLD SELECTOR

HIDE CONTROL CODES ON/OFF

31. CHANGE DIM

8. REPEAT DIM SELECTOR

9. VERTICAL RULE

10. BELL CODE

11. TAPE FEED

32. LINE INSERT
CHARACTER REMOVE

33. LINE REMOVE
WORD REMOVE

12. UPPER RAIL

13. PAPER FEED

14. THIN SPACE

15. HERE IS MULTIPLE CODES

34. PARAGRAPH REMOVE
SENTENCE REMOVE

35. SET/SKIP

36. REMOVE TO CURSOR
CURSOR UP

16. ELEVATE

37. SET/TAB

17. EN LEADER

38. CURSOR LEFT

18. EM LEADER

19. RETURN

20. SHIFT

21. UNSHIFT

22. QUAD LEFT

23. LOWER RAIL

24. QUAD CENTER
RUBOUT

39. END OF TEXT (ETX) 44. STRIP OPTION SELECTOR

HOME CURSOR

45. CURSOR I/O SELECTOR

40. CURSOR EXTREME RIGHT MEMORY OUTPUT SELECTOR
CURSOR STEP RIGHT

46. INPUT DEVICE SELECTOR

41. CLEAR LINE INPUT START/STOP

ROLL DOWN

47. OUTPUT DEVICE SELECTOR

42. CLEAR REST OF MEMORY OUTPUT START/STOP
CURSOR DOWN

48. ADD OPTION SELECTOR

43. HOME MEMORY

ROLLUP 49. FEED TAPE SPACING

Composition Input 79

Some VDT’s do not display some typographical characters and commands.
Here are some of those symbols displayed by one VDT system.

Quad Right (QR)

Quad Center (QC)

4->

Quad Left (QL)

«—

Upper Rail (UR)

Lower Rail (LR)

Em Space (EM)

En Space (EN)

□

Em Leader (EM LDR)

• •

En Leader (EN LDR)

•

Vertical Rule (VR)

Tape Feed (TAPE)

t f

Paper Feed (PAPR FEED)

P F

Return (RWT)

Elevate (ELEV)

•u

Shift (SHFT)

Unshift (UNS)

Thin Space (THIN)

Add Thin Space (ADT)

III

Bell (BELL)

Rub Out (RUB)

• • •

The Hendrix 5200 video display.

80 Composition Input

Here are the control keys of the Mergenthaler
CorRecTerm video display terminal. The cluster of
keys at right controls editing functions; that at left
controls cursor position; and that above controls in¬
put and output.

Control of a VDT

The cursor, displayed by alternating the background of a character between
black on white (BOW) and white on black (WOB), indicates the position on
the screen at which a function is to be performed. It blinks to take the
operator’s attention to that point on the screen. A character entered from the
keyboard appears on the screen exactly where the cursor is currently located,
then the cursor automatically moves to the next character position.

Return Designed to be used for test entry and its function depends on which
mode is operating, as follows:

FORMAT MODE — RETURN is used to return the cursor to the
beginning of the next line on the display.

JUST or NW MODE — RETURN inserts customer specified line delimiter
at the end of a line and moves the cursor to the beginning of the next line.

UNJUST MODE — RETURN inserts customer specified paragraph
delimiter at the end of a paragraph and moves the cursor to the beginning of
the next line. In this mode, all cursor-returns at the end of lines are done
automatically.

CR (CURSOR RETURN) — M oves the cursor to the beginning of the next
line. CR was designed to be used for text editing and to be independent of the
mode of operation.

HOME —- Moves the cursor to the upper left-hand corner position on the
screen called “Home.”

By depressing any one of the arrow keys, the cursor is moved one position at
a time in the direction of the arrow and will repeat in that direction when
held down.

END OF TEXT — Moves the cursor to the exact end of text which is
defined as the point where the operator can start entering new characters in
order to append previously entered text. If the end of text position is not on
screen, ROLL-UP’S will be automatically generated to bring it into view
where the ETX function can be performed.

RIGHT — Moves the cursor to the right side of the line the cursor is located
in.

Composition Input 81

Editing

Many advanced editing controls are often provided so that the operator can
rapidly and effectively edit text. For the convenience of the operator and to
help speed up the editing process, a WORD, SENTENCE or
PARAGRAPH can be removed in their entirety with just a few key strokes,
in addition to removing single characters at a time.

OVERSTRIKE is the normal keyboard mode of operation where each
character entered from the keyboard replaces the character in the location of
the cursor. The cursor will step right and no other character on the screen
will be disturbed.

The operator should select the INSERT mode (INSERT light ON) when he
desires to insert a missing character between two characters. A character
entered in this mode is inserted at the position of the cursor, pushing the
character previously there to the right one position.

The keyboard is automatically switched from the INSERT mode to the

OVERSTRIKE mode when the cursor is moved by any of the cursor control
keys.

CHAR (CHARACTER) removes the character in the cursor position,
closing text up. Keeping CHAR depressed continuously removes characters
by pulling text in from the remainder of the line.

CCASE (CHANGE CASE) changes the case of the character under the
cursor from upper case (A,B,C) to lower case (a,b,c) or vice versa. Especially
helpful when an input tape has missing unshift or shift codes.

BOLD displays characters as they are entered from the keyboard as black on
white background instead of the normal white on black. BOLD keyboard
entry (BOLD light ON), is useful when an operator wishes to distinguish
bold face characters from the remaining text. This saves the operator from
having to enter upper and lower rail codes.

CBOLD (CHANGE BOLD) reverses the display of the character under the
cursor from NORMAL to BOLD or from BOLD to NORMAL, depending
on the initial state of the character. Useful for going back over text and
changing characters to or from BOLD face.

DIM displays characters entered from the keyboard as either: gray
characters on black background for DIM/NORMAL ENTRY (DIM light
ON only); or the reverse, displayed as black characters on a gray
background for both DIM/BOLD entry (DIM and BOLD lights ON).
Useful to distinguish italic or other type characters from the rest of the
displayed text.

CDIM (CHANGE DIM) reverses the display of the character under the
cursor from either: NORMAL to DIM/NORMAL or DIM/NORMAL to
NORMAL; or from BOLD to DIM/BOLD or DIM/BOLD to BOLD.
Useful for going back over text and changing NORMAL and BOLD
characters to or from their DIM configurations.

82 Composition Input

WORD removes the word in which the cursor is located, closing text up.

SENT (SENTENCE) removes the sentence containing the cursor, closing
text up.

PARA (PARAGRAPH) removes the paragraph containing the cursor,
closing text up.

LIN (LINE INSERT) moves all text from the line the cursor is located in
down one line, leaving a blank line.

LRM (LINE REMOVE) removes the line containing the cursor, moving the
remaining text on the screen up one line to fill the space.

CLL (CLEARLINE) erases the line containing the cursor, leaving a blank
line in its place.

Rolls text on the screen up one line at a time “forward” through main
memory, until it stops at the bottom of memory. RU can be Roll up, Roll
Down the beginning of the story. Rolls text on the screen down one line at a
time “backward” through main memory, until it stops at the top of memory.

«Ihe Hergenthaler Coi
systei for proofing, cc
editing text ~ prior
typesetting systeis.I
Tie fferoenthaler C r

m <>ri

'^sistifiQ of > , v

level paper tape
and a combination CRT
p iput (TTS) keyboard.

* nd aesthetically, these

ideally suitted to
or plant.[
er« serves these
typesetting features:!

5 fast, one-tan tethtd of
ig and tape terging of
ections.1

Remits easy insertion of fwction
or paraaeter codes to coaputer
input tape.!

..reader
•».e that
■*nd

The Mergenthaler CorRecTerm. The unit at right
reads in tapes and perforates new ones.

Composition Input 83

Operation

Keystrokes
per Hour

Ems
Approx,
per Hour

-Equiv.

Typewritten

Words/Min.

1. Manual Linecaster

7,000

3,500

2. Monotype Keyboard

10,000

5,000

3. TeleTypesetter Keyboard

10,400

5,200

(Justified Lines)

4. Typewriter Keyboard

18,000

' 9,000

(Non-justified Tape)

5. Typewriter

23,800

11,900

One frequently asked question is that of expected keyboading speeds. This is
one of the reasons for the creation of the National Composition Association
Production Measurement Committee which is finalizing its report on
keyboard production standards. It will be interesting to compare some of the
conclusions of the Production Measurement activities against your own
findings. While these standards are not based on a controlled study and do
not indicate the specific type of equipment used, they do offer some
interesting information.

It is obvious that a manual linecaster offers some mechanical limitations
which affect the productivity of the equipment. The Monotype and
TeleTypesetter keyboards also offer some mechanical limitations along with
end-of-line decisions which can hamper total productivity. In these
standards, the only apparent difference between typewriter keyboard (non-
justified tape) and straight typewriter is the typing skill of the operator. This
confirms some of the preliminary conclusions drawn from the NCA study.
In fact, we have suggested that all operators be given a typing test in order to
ascertain their production potential. A copy of the typing test is available
from NCA/PIA headquarters.

84 Composition Input

7. Optical character recognition (OCR) systems

Before the typesetting and printing process can begin, manuscript material
must be converted. This conversion process embraces two very important
functions:

-it converts the material to a form compatible to the typesetter
being used. In practice, this usually means the creation of a
punched paper tape in the appropriate 6 level TTS format.

-it adds the typesetter control instructions, so that the finished
textual material is of the desired format, and appears in the
right font and size.

There are conversion processes that accomplish both steps by a single
conversion operation, while other processes require two separate steps to be
taken.

The actual choice of the conversion process will depend to a great degree on
the form of the manuscript, or document. Where the manuscript is in the
form of a typed page, it may be possible to read by pages directly by using an
OCR system. Or, the pages may be converted to tape by using keyboard
devices. Where the manuscript is in the form of machine readable tape, it
may be possible to use this tape to create useful typesetter input.

Extracting information from a document falls under the general heading of
character recognition. This idea was first developed by the data processing
world in an attempt to eliminate the input conversion step. It was found that
many documents were generated by a machine to impart information to a
human, and that this same information was subsequently required by a
machine. A primary example of this problem is the utility bill or credit card
bill. A slightly different problem exists where it is necessary to give unique
identity to a document, such as a check. In all these cases, the general
principles of character recognition were promoted so that it was not
necessary to use a keypunch operation to get this material back into machine
readable form. As the volume of such keypunching increased, the value of
character recognition equipment also increased.

A number of machines are now available to address certain specific tasks,
such as reading journal tapes from adding machines, reading sales slips

Composition Input 85

embossed by credit cards, reading hand marked test scores, and so on. A
number of these machines have important constraints on the number of
characters they can recognize, the type of material upon which the
characters are printed, the size of the document they can accept, etc.

In general, two basic recognition schemes are being used today (1) reading

the character by magnetic means, or (2) reading the character by optical
means.

The most widely used magnetic method of character recognition is called
MICR (Magnetic Ink Character Recognition). In this scheme, the
characters are printed using special magnetic ink. The characters, due to
their distinctive shape, have unique magnetic properties, and it is possible to
read these characters by using a special magnetic reader. The character set is
limited, consisting only of the numbers 0-9 plus 4 special characters.

The banks are the largest users of this concept. Printed checks usually
contain the name of the account holder (be it an individual or a company)
and the account number using MICR characters. In this way, the check can
be machine read to establish the account number.

The associated code is a machine readable code associated with a human
readable character. In this way, the information is conveyed to the human
reader by conventional characters, while the machine reads the associated
code. A great deal could be said about the applications of the different
techniques, but our primary purpose is to simply expose the concept.

Datatype uses an IBM Selectric typewriter equipped with a special ball. It is
also possible to modify a line printer to produce this font, but the primary
value of the idea is the ability to use a standard office typewriter as a data
recorder. Once the page is produced using the Datatype code, it can be read
by a special page reader made by Datatype. The reader scans the page, a line
at a time, and converts the information to a punched tape or magnetic tape.
The machine readable code located under the human readable character is
similar in principle to the Semagraph, and shares the drawback of being
limited in the number of codes available. Datatype can print 88 codes, and
since some of these codes are related to control, they do not print a human
readable character. This has the effect of reducing the character set of the
typewriter. The Datatype page reader is in the $10,000 class, and the reader
is used in conjunction with a minicomputer and tape punch or mag tape drive
to form the entire system.

The Potter approach uses bits printed both above and below the human
readable characters. Potter has based the system on the use of cards, which
can be printed by means of a special typewriter (either Selectric or type bar)
or a special line printer. The value of the system seems to be in the ability to
sort and collate cards. The card reader is available with interfaces to permit
connection directly to a computer, thus the need to convert to a tape is
eliminated. The reader is rather expensive, being in the $20,000 range. The
character set is limited to 64 codes.

86 Composition Input

Dual Image uses a full 8 bit code pattern printed below the human readable
character. The images are generated by a special printer, which can print a
character set of 128 different symbols. In addition, the 8 bit code pattern can
furnish any of the 256 different possible codes. The printer can also produce
two character mnemonics using a single character space. The printer
produces a strip of tape, which is then read by a very simple reader in the
$1,000 price range. Interfaces are available to connect directly to computers,
so there is no need to convert to a different machine readable media. The use
of 8 bits in the code pattern also permits the printed code to be in machine '
compatible format, so that no code conversion is required.

Mark Sense requires the use of a special mark placed in a particular zone on
the document. The mark is usually made by a soft lead pencil, and the reader
merely determines the absence of or presence of this mark. The concept is
widely used in scoring tests, where the answer may be “true or false” or
possibly one choice in five. The business world uses a similar scheme, but it is
based on a numeric only technique where the mark is placed in a one out of
10 basis. Mark sense readers are marketed by IBM, NCR, and Hewlett-
Packard (to name a few) and start as low as $3000.

OCR, itself, recognizes the character by its shape, similar to the way the
human eye reads. OCR readers are classified by their ability to read
different font styles or character shapes (single or multifont), and by the
number of different characters that can be recognized (sometimes referred to
as the “vocabulary” of the reader).

Single font readers are manufactured to read a specific font style, such as the
USASI OCR-A font, or the European ISO OCR-B font.

Since these fonts are stylized, they must be generated by typewriters or line
printers that are so equipped. In general, the spacing between characters, the
spacing between lines, and the reference to a given side of the page must be
rather precise. OCR readers are frequently sensitive to paper material, since
the “reflectivity” of the paper vs the reflectivity of the image must be
consistent.

Multi font readers have the ability to be programmed to read any number of
different font styles. They usually have a fairly large vocabulary, and can
read upper and lower case characters, the number set, punctuation marks,
plus some special characters. Some machines can even read carefully
executed handwriting. In the main, these machines can accept material that
has been produced on standard office typewriters or line printers. Obviously
the more capability of the machine the higher its cost, and large vocabulary-
multi font readers are very expensive.

As with single font OCR readers, the multi font system reads the input page
and produces a machine readable tape, either magnetic or paper. Since a
computer is included as an integral part of the system, it is possible to do
some formatting of the output material. The nature of the formatting, and
its extent, is a function of the computer program.

Composition Input 87

Document readers can generally read one-to-five lines of information from a
paper coupon, stub card, or similar document. Most document readers can
handle documents ranging in size from 2x4 inches to about 4x8 inches.

Document readers are widely used in reading turn-around forms such as
statement remittance stubs where the printed output from a computer later
becomes input to the computer.

Page readers are generally designed to read large and variable amounts of
alphanumeric information typed or printed in normal page format. Most
page readers accept sheets from 8 b x 11 to 12 x 14 inches in size. Page
readers also have the capability of reading somewhat smaller sheets and
continuous fan-fold or rolled sheets printed by computers or by specially
equipped typewriters.

Bar-code readers sense marks that are used in combinatorial form to
indicate data. The type of marks used varies, but in most cases the marks
cannot be formed by hand and are not easily readable by humans.

Usually, special devices are required to produce the bar code imprinting. Bar
codes also suffer because the code and the character occupy a good deal of
space.

Character readers are the upper-class of optical readers. They translate
human-readable characters into machine-readable form. Many specialized
fonts have been developed to simplify the character recognition logic and
hence lower the price.

A document reader reads documents of less than standard letter size (8.5 x
11 inches). A page reader reads at least letter-size documents and usually
larger ones. Another way of distinguishing between document and page
readers is that document readers generally read one or two lines per
document, while page readers can read many lines from each document.
Single font means that the reader can be equipped to read one typeface only.
Multiple font means that the reader can be equipped to read several
typefaces but only one at a time; switching between type faces can be a
manual or programmed feature. Multi-font means that the reader can read
multiple typefaces intermixed; this is the most sophisticated and expensive
type of optical reader. Journal tape is the rolls of tape used by adding
machines and cash registers.

Now that we have defined the types of devices we re talking about, let’s
discuss their application.

Who Uses Optical Readers?

Mark readers are used principally for data collection and for entry of limited
amounts of data on previously punched cards.

Bar-code readers and character readers have many applications. The
principal ones at present are the reading of slips imprinted with a credit card,
processing of turnaround documents, and sorting of the U.S. Mail.

88 Composition Input

Various manufacturers estimate that any installation having anywhere from
7 to 12 or more keyboards can profitably make use of a character reader.
Where do the savings come from to pay for this expensive beast?

One place is the lessened cost of labor. Since manual input for the character
readers is typically prepared on a typewriter, the hourly wage rate is
generally lower than for keyboard operators, while the output is higher and
the rate of errors is lower. The ease with which errors can be corrected when
preparing typewritten documents contributes to the speed in comparison
with keypunching. One' user estimates that about 10 percent of the
documents processed at his installation contain errors detected and
corrected by the operator. Some readers contain special facilities for
recognizing a character skip symbol or strike-throughs to further ease
correction of errors detected by the typist.

Most optical character recognition systems consist of four basic units.

Document Transport Unit
Reading Unit
Recognition Unit
Control Unit

The transport unit moves documents past a scanner that converts the
characters on the document into electrical signals that are then analyzed and
recognized by the recognition unit. The recognition unit matches patterns or
representations received from the scanner against stored reference patterns.

The transport moves the documents from an input hopper or feed roll past
one or more scanning units to one or more output stackers. In certain
equipment the documents are read while still moving, but in most cases, the
document is stopped and read. Document transports employ combinations
of vacuum, air blast, and friction to separate and feed individual documents,
while belts and rollers are used to transport the documents past the scanning
unit.

The speed of most OCR systems is limited by the speed of the document
transport. The scanning unit determines the speed of the OCR unit when the
amount of data per document is large.

Functional features or characteristics that effect the complexity of the
document transport are:

Detection of double documents and jams
Detection and correction of document skew

The scanning unit converts the printed information on the document into
electrical signals that will enable the recognition unit to recognize the printed
characters. The five distinct methods currently in use for converting optical
signals into electrical signals are:

1. The rotating disc scanner uses a high quality lens system to project light
reflected off the document onto a rapidly rotating disc. The rotating disc has

Composition Input 89

apertures extending from the center of the disc to its periphery. Behind the
character image area on the disc is a fixed plate containing a single aperture.
This aperture is so oriented that each aperture on the rotating disc
successively intersects along the entire length of the fixed aperture as the disc
rotates.

The rotating disc scanner reads one character at a time. Movement from one
character to the next character or from line to line is accomplished by
repositioning the lens system or by moving the document. Therefore, this
type of scanner is relatively slow in comparison to other scanning methods
mentioned.

The advantage of the rotating disc approach is that it is relatively simple,
permits paper to be exposed to ordinary light (actually the more light the
better), requires only one or, at most, a few photocells, and permits
adjustment for different background colors by varying the threshold voltage.

The disadvantages of this method are that high-speed discs are noisy and
difficult to manufacture and the throughput rate of the system is limited by
the disc revoltion speed. It appears that about 400 characters per second is
the upper limit for OCR systems employing this approach. Therefore, the
rotating disc is being replaced by faster methods of scanning for a number of
applications.

90 Composition Input

2. The flying spot scanning method uses a cathode ray tube (CRT) to
generate a small (spot) of light that is projected onto the document being
read via a lens system. The document must be located in a lightproof
compartment where the reflected light can be picked up by one or more
photomultipliers.

The CRT light beam is swept across the character in a raster-type screen by
the CRT control logic. The beam can be moved very rapidly by the control
unit to any location on the document. This ability enables a flying spot
scanner to locate a line of print anywhere on the document without having to
read the entire document. This positioning capability may also be used to
follow the lines of a character, thus making it particularly adept at reading
multiple font and hand printing.

3. In a photocell scanning system, a high intensity light source illiminates the
document which is in motion and the reflected image is focused onto a
grouping of photocells or light pipes which feed the photocells. The grouping
of photocells can either be vertical, relative to the character being read, or it
can be a parallel array of photocells.

In the vertical grouping of photocells, each character is sampled as it moves
from left to right. The use of a vertical grouping of photocells speeds up
scanning operations by simultaneously sampling a number of points which,
when combined, add up to a complete vertical slice of the character. The
electrical signals generated by each of the photocells are then coverted into a
binary mode and each slice is stored in shift registers until the entire
character is sampled.

Composition Input 91

The parallel- array approach looks at all points of a character
simultaneously. The speeds of each method are comparable; however, in the
parallel or full photocell array approach it is possible to measure analog
information completely. This capability enables different shades of black
and white to be read and thus provides a greater probability of recognizing a
smudged or dirtied character.

4. The vidicon technique projects the characters to be scanned onto a vidicon
television camera tube. The vidicon tube is instantaneously exposed to the
characters (in camera fashion) by either flashing a light (flash tube) on the
document when the characters are ready to be read, or the document is
constantly illuminated with a strong light and a high speed electro¬
mechanical shutter is used to “snap the picture.” The image on the face of
the tube is then scanned by an electron beam which generates an electrical
analog signal. The resulting signals are quantitized to digitally indicate black
or white.

The vidicon scanner can store a group of about 45 characters on the face of
the tube and, therefore, documents containing this number or fewer
characters do not have to be moved during the scanning operation. With the
development of much higher resolution vidicon tubes, it would be possible to
store the entire document and eliminate mechanical movement completely
during the scanning operation. Vidicon scanners are presently classified as
medium speed (500) characters per second.

After the scanning has taken place, an electrical representation of the
character is transmitted to the recognition unit which then identifies the
character. With some systems there is an intermediate prerecognition step in
which undesirable electronic “noise” caused by white spots on black ink,
dirt, or inadvertent ink spots is reduced. The value of this technique is still in
dispute. The most common types of recognition units currently in use are:

1. The use of optical masks is one of the earliest recognition techniques. It is
based on the use of one or more photographic masks for each character. An
attempt is made to measure how well the character projected matches with
the mask.

92 Composition Input

Photocells behind the mask measure the total light passing through tlje
mask. Ideally, no light should pass through the mask if it matches the
character being identified. In practice the match is usually not precise
enough to blank out all the light so a threshold value is established as a
tolerance.

This technique has the ability to identify a full alpha-numeric character set,
however, small differences in character shape may cause character
identification errors and high reject rates. No known commercial OCR
systems are currently using this method. The concept has the potential for
providing low-cost OCR systems if the input data can be closely controlled.

2. Matrix matching is a widely used recognition technique which stores
electrical signals received from the scanner in a digital register that is
connected to a series of resistor matrices. Each matrix represents a single
reference character. Each resistor matrix is connected to a second digital
register which contains a voltage representation of the character. The
voltage of the scanned character is compared with the second digital register
and the resistor matrix.

Recognition is based on the comparison of the voltage representations in the
two-shift registers. This recognition technique is well developed and can
handle a complete alpha-numeric character set and is easily modified to
identify characters from several type fonts.

3. Analog waveform matching is a recognition method that has been used for
some time, particularly in the magnetic character reader used by the banking
industry. This method is based on the principle that each of certain
characters passing under a read head will produce a unique voltage
waveform as a function of time. Characters are identified by matching their
waveforms against reference waveforms.

The major disadvantage of this technique is that only a small number of
characters have easily identifiable waveforms, thus limiting this application
to the reading of only numerics plus a few special symbols. Machines using
this technique have reading speeds of approximately 500 characters per
second.

4. Frequency analysis is a digital recognition technique developed for fonts
using vertical lines. The CMC-7 font printed ith magnetic ink, is the most
widely used example of this technique. The width of the gaps between the
vertical lines of a character form a code that is unique to the character.
Characters can be identified by comparing the frequency and number of the
wide and narrow gaps with the stored codes for each alpha-numeric
character. The advantages of this technique include the ability to handle a
full character set.

5. Stroke analysis or feature analysis is a recognition technique based upon
the differentiation of characters by the number and position of vertical and
horizontal strokes or lines. The formation of the unknown character is
matched by a special purpose computer against stored truth tables
representing each reference character. The capability of this technique has

Composition Input 93

been increased to the point where hand-printed numerics, and several fonts
can now be recognized.

The systems control unit performs data editing and formating, identifies and
interprets various formats for different documents, sequences some systems
operations, and provides the interface required to record data on an output
device such as magnetic tape or punched card. The systems control unit may
be a special or general purpose computer, or a plugboard.

In most of the early optical characters readers, systems control functions
were performed with plugboards that greatly limited the system’s flexibility
and data processing capabilities. While some readers still use a plugboard,
particularly mark sense and document readers, most current OCR systems
and those expected in the future use a combination of systems software
capability and a general purpose computer for control functions. The
computer can be a mini-computer supplied as an integral part of the system
or the reader can be operated on-line to a larger data processing system.

OCR devices with computer logic capability and specialized software are
performing many sophisticated data processing functions such as validation
of self-checking numbers, reconstruction of missing digits, and identification
of characters by context analysis.

There is not too much to say about recognition of marks by a mark reader.
Typically, they are diagonal slashes made in a preprinted box or outline.
Care must be taken when erasing because if the paper is roughened too
much, it will have a low reflectance and make the reader conclude that the
roughened area is a mark. In general, more care must be taken with erasures
than with the older conductive mark-sense technique.

The style of the type face printed is called a font. The group of symbols that
the reader will recognize is referred to as the character set. Note that a
particular reader may not read all the symbols of a particular font. The usual
situation is that only the numerics are recognized of a font that also contains
alphabetic letters. Occasionally, a larger set of characters can be recognized
than are in the font proper.

There are many fonts that are used today in addition to traditional printing
type styles. A few of the more sophisticated readers recognize printing type
styles and in addition to or in place of the OVR printing type styles in
addition to or in place of the OCR fonts.

The more commonly used OCR fonts include:

• NCR NOF (Numeric Optical Font) — This is a numeric font, usually
imprinted by an adding machine or cash register. It is widely used in retail
applications.

• ANS I and IV — Previously called OCR A and C, these fonts were
developed by the American National Standards Institute and are likely to
become the most widely used OCR fonts. The two represent two sizes of the
same typeface. For perspective, the sizes are roughly equivalent to 10 point
(I or A) and 14 point (IV or C) type. Imprinting devices abound, including

94 Composition Input

the IBM Selectric Typewriter, most other major brands of electric
typewriters, and some line printers.

• OCR A and C — See ANS I and IV, above.

• ISO B Popular in Europe, this font is now under consideration for
standardization in the United States. It differs considerably from the ANS
standard and is characterized by being closer to conventional type faces than
the ANS fonts. Exponents of the ISO B font are concerned about readability

by people, even though it is more difficult to build the machine recognition
logic to handle it.

• Farrington 7B, 12F, and 12L — Another popular group of OCR fonts, due
to Farrington’s early appearance on the OVR scene. The three codes have
somewhat similar shapes but differ in size and character set. The 7B and 12F
are numeric fonts, while the 12F is alphabetic only. The 7B is much larger

than the 12 F/F. Imprinting is normally done by a special typewriter or a
credit card embosser.

• IBM 1428 This is an alphanumeric font associated with the IBM 1428

Optical Reader and imprinted by an IBM 1402 Fine Printer or IBM
Selectric Typewriter.

• IBM 407 — A font produced by the widely used IBM 407 Accounting
machine.

• E-13B — This is not properly an OCR font. It was developed by and for

banks prior to the development of OCR. It is a highly stylized numeric font

intended for printing an magnetic ink to facilitate the sorting and processing

of bank checks. It has not caught on anywhere else, but most banks use it.

Some optical readers can read this font, which enhances their suitability for
banks converting to OCR.

All fonts, both numeric and alphanumeric, usually include a few special

symbols for control purposes. The OCR A and C fonts include a full array of
punctuation symbols as well.

The scanning technique is the method for optically converting the printed
images to electrical signals. Some sort of photosensitive device, photocell or
phototransistor, is used to sense the light reflected from the document. For
bar-code and character readers, additional components are required to scan
portions of the code or character in proper order so that the features, and
thus the character, can be identified. The scanning components can be an
array of photo devices, a mechanical disc, or a flying spot (CRT). The
photo-device array is used by most bar-code readers. Size normally
interferes with using it for scanning characters, but note that REI does quite
nicely with it. The mechanical disc technique employs a rotating disc with a
slit in it to project a beam of light over the character in a predetermined
order. The Hying spot scanner uses an electron beam that is moved within a
CRT to generate a spot of light, thus providing, potentially anyway, a much
faster scan rate. The Hying spot scanner is very adaptable to reading multiple
lines, while the mechanical disc scanning technique requires either an

incremental document transport or an elaborate system of mirrors to scan
multiple lines.

Composition Input 95

Once the printed image has been translated into electrical signals, the
recognition logic interprets these signals as a particular character.

Three principal recognition techniques are used for character interpretation:
matrix matching, stroke analysis, and curve tracing. Matrix matching
involves comparing the matrix of signals caused by the reflecting (paper) and
non-reflecting (printed character) portions of the character with a set of
signals for each character until a match is found. Typically, the closest
match within prescribed limits is identified, because a perfect set of signals is
most unusual due to variations in print and paper quality. (The stroke
analysis method is somewhat similar on a much simpler basis.) Readers
employing this technique usually are reading a highly stylized font
specifically designed for the technique. Curve tracing logic actually traces
out the outline of the character to derive a set of signals for analysis. The
curve tracing technique is adaptable to variations in character size and
orientation, making it a good choice for interpreting hand-printed
characters. However, breaks in the printed character tend to affect this
method more than the matrix matching method.

The purpose of the optical reader is to generate data in a typesetter -
readable form. Magnetic tape, punched cards, and punched tape are
conventional computer-readable forms. De facto standards, now official,
have been established by IBM for magnetic tape and punched cards!
Teletype has done essentially the same for punched tape. Exceptions that are
relatively new on the market are magnetic tape cassettes and 96-column
cards — but neither of these have had much impact on the optical reader
market, for the simple reason that they haven’t had much impact on
computers as yet.

Readers that handle one size of documents are easy to rate for performance
because it is predictable. In a similar manner, a journal tape reader typically
transports the tape at a fixed rate, with predictable performance. Readers
that handle different sizes of documents and variable-size data fields are not
as conducive to having their performance stated in simple terms.

Three ways of measuring the performance of optical readers are documents
per minute, lines per minute, and characters per second. The documents-per-
minute rating is usually most applicable to mark and bar-code readers, as
well as to character readers that read only one or two lines. The lines-per-
minute rating is usually most meaningful for journal tape readers. The
instantaneous character scanning rate in characters per second is probably

the most meaningful single measure for character readers that read whole
pages of text.

Careful evaluation of timing information, which often becomes quite
complex, is necessary to accurately predict the performance of the more
sophisticated character readers. The size of the document, the amount and
location of data on the document, and processing of the data read can all
affect the rate at which documents proceed through the reader. One-line
units can be affected by other activities of the computer, if running in a
multiprogramming environment, or by poor programming of input/output
functions.

Reject rates of 0.25 to 0.5 percent cause some users to become startled and

96 Composition Input

disillusioned when they see 30 or 50 percent of the documents going into the
reject pocket.

There are three principal types of errors of concern to users of optical
character readers: ambiguous characters, invalid data, and documents in
poor condition.

Ambiguous characters are those for which the reader cannot make a
decision about what character each should be. There can be many reasons.
Typical ones include broken or poorly formed characters and dirt or other
marks that are picked up by the reader. Handling of this situation varies
with the reader and with programming. Many readers automatically rescan
an ambiguous character. Some substitute a standard character for all
unreadable characters and continue. Others display the character on a CRT
screen for operator determination; sometimes adjacent data is also displayed
to give the operator more context for making the decision. Printing quality
and paper quality can drastically affect the incidence of this type of error.

OCR users quickly learned that the inclusion of checks in the data was
extremely useful for insuring the maintenance of an adequate throughput
level by reducing the number of rejected documents. This technique most
frequently takes the form of repeated data fields, particularly for numeric
entries. The technique is applicable only if the data can be processed and the
actions of the reader controlled on the basis of the result.

Another commonly employed check is the check digit. The digits of a
numeric field are manipulated, and there are several standard formulas, to
generate a check digit. This digit is included in the input. The reader or
associated processor generates another check digit while reading and
compares it to the one read in. Failure of this check normally causes the
document to be rejected.

Documents that have extraneous items on them (such as stamps) or that
have been badly mutilated can cause misfeeding and/or jams. The typical
character reader is far less susceptible to this kind of jam than the average
card reader, but people have been conditioned not to fold, spindle, or
mutilate punched cards.

To permit processing of the manuscript by an OCR system, the typed pages
must satisfy the criteria of the OCR reader. Some of these constraints are:

-the type style must be one the OCR reader "knows”

-the character set should not exceed the character library

-the paper or document material must be acceptable to the reader

-the line length, and line placement, must be appropriate for the reader

The first step is to read the character, and this is done by using scanning
techniques. A single point of light is moved over the character image, and the
light and dark portions of the scanning pattern are fixed on a matrix. The
dark portions of the image represent dots on the matrix, and it is then
possible to assign X and Y co-ordinate points to the dots. Thus, the
character image is converted to a dot pattern, and each of the dots can be
represented electronically as a result of the X-Y co-ordinates.

Composition Input 97

OCR readers scan several hundred lines in each of the two planes to achieve
fairly high levels of resolution. The process of reducing the image shape to a
dot pattern on a matrix is sometimes called "digitizing”, or converting the
image to digital representation.

Scanners are either mechanical or electronic, with the mechanical scanners
controlling the movement of the light point by means of mirrors. Some
mechanical scanners position the document on a drum, and rotate the
document to derive one plane, and a mirror establishes the other plane.
Electronic scanners use CRT principles and control the light point by
deflection plates.

Once the character image has been digitized, it is then passed on to the next
stage, where the character being read is compared to a library of all known
image patterns. This is called correlation. When a reasonable comparison is
achieved, the character is identified.

The identified character is now passed on to the format and control element,
where the material is organized for proper recording on the output tape. The
control and format section also takes care of such things as discarding
characters that had been deleted. Some systems also utilize a small TV
screen, so that if the reader scans a character, but can not find a comparison
in the character library, the unknown character is displayed for the operator
to see. The operator can then interpret this image, insert the correct
character back into the system via a keyboard, or take other appropriate
action.

The output of OCR systems is either magnetic tape or punched paper tape.
The higher speed machines use magnetic tape because of the higher

recording speeds, although paper tape can also be used if the speed sacrifice
is not important.

OCR has developed very slowly as evidenced by the fact there are currently
only about 1,500 OCR installations versus about 80,000 computer
installations, even though the first commercial OCR system was installed in
January 1956, only three years after IBM shipped its first commercial
computer. The primary reasons for slow growth of OCR were the lack of
flexibility of the first generation OCR systems and limited view of the
potential OCR market by the industry and potential users. Until recently,
almost all OCR devices were limited to reading short vertical or horizontal
dashes (mark sense), or highly stylized fonts designed primarily for OCR
reading. The lack of flexibility of earlier OCR systems has been remedied
and most OCR systems introduced since 1968 and those scheduled to be
introduced in the near future are capable of reading a number of different
type fonts, including hand-printed numerics plus a few symbols. In addition,
the software programming has been greatly improved giving the systems
considerable editing and formatting flexibility.

Optical character recognition systems can be classified by the complexity of
the font selection which the machine can handle and by the physical
characteristics of the media presented for reading.

98 Composition Input

OCR systems fit into the following categories when classified by font
selection:

1. Optical Mark Readers

2. Stylized Font Readers

3. Multifont Readers

Optical mark readers are an improvement in the mark sensing techniques
where the location of graphite pencil marks was determined by measuring
the electrical conductivity of the pencil mark. Now, most mark reading is
done optically. At one time, optical mark readers were used primarily in
scoring tests and questionnaires and in survey applications. Today, they are
also used as data acquisition devices in payroll, inventory control, meter
reading, and a number of similar applications.

The scanning systems of optical mark readers and optical character readers
are basically the same, but their recognition systems differ greatly. Both
systems use a set of photocells to detect a drop in reflected light caused by a
mark or character on the paper. The recognition system for an optical
character reader is considerably more complex than that of an optical mark
reader. An optical mark reader recognizes only the drop in light and
determines the value of the mark by noting its position in relation to some
reference point such as the beginning of a line. An optical character reader
notes not only the gross drop in light but the coordinates on a two
dimensional grid at which the drop occurred.

There are approximately 50 major type fonts used by computer printers and
typewriters. This large number of different fonts has presented a great
restraint on the development of optical character recognition systems. In
order to control data input, the initial manufacturers of OCR equipment
created special stylized fonts that were compatible with the design of their
own equipment. This lack of standardization during the past decade has
severely limited the growth and acceptance of OCR as an input technique.

A number of type fonts have been developed over the past several years for

machine reading. The USA Standards Institute has recommended a type

ont that is generally suitable for typewriters and is also suitable for OCR.

The USASI OCR-A font is the most common font, and is read by almost all
OCR systems.

Composition Input 99

The following pages are excerpted from the Olivetti
manual on Optical Character Recognition fonts and
keyboard arrangements.

This booklet illustrates the most common Editor 2 key¬
board arrangements for use in Optical Character Re¬
cognition applications.

When ordering an OCR Editor 2*» the following questions
must be answered:

1. Uhat character set is the Optical Reader pro¬
grammed to recognize?

2. Uhat Keyboard will the application require?

3. Uhat Keyboard modifications will be required
due to special symbols such as the character
delete?

M. Uhat size carriage will be required to accommodate
the form size?

5. Will the forms be continuous or cut sheets?

b.* Uill carbon or fabric ribbon impressions be re¬
quired?

Use the following information to obtain answers to the
above questions.

Character Set

A keyboard number is located in the space bar on each key¬
board diagram. This numberi used when ordering** describes
both the keyboard arrangement and the character set. For
example-* keyboard 2-^03-! indicates a 10-Key Keypunch
cluster keyboard arrangement and the USA Standard Character
Set •

A type sample precedes the diagrams of the keyboards avail¬
able with each character set. The Optical Reader will de¬
termine the character set that is required.

Keyboard

After determining the character set** the keyboard should
be selected-

The keyboard should be chosen to meet the needs of each
application. A description of each keyboard has been in¬
cluded for this purpose.

100 Composition Input

Keyboard Modifications

The special symbols will also be determined by the Optical
Reader. The most frequently used character combinations
are listed on page 3 • Others are available upon request.

Any keyboard may be modified to include special symbols*!
when required.

Carriage Length

The size of the form will indicate the length of the
carriage to order.

Pin Feed Platen

When continuous forms are used*, a pin feed platen may

be installed on the Editor 3. Refer to the current

Technical Information Bulletin I TIB > for the ordering

procedure*, prices and installation data. Additional

information is contained in the Features and Attachments
Booklet.

The maximum form widths

Carriage length

13"

17"

for each pin feed platen are:

Maximum form width

11 7/A"

IS 7/6"

Ribbon

Most Optical Readers recognize
carbon ribbons.

impressions produced by

Composition Input 101

USASI * OCR type style and keyboards

THIS IS A SAMPLE OF THE UNITED STATES OF AMERICA
STANDARDS INSTITUTE-, USASI*-, OPTICAL SCANNING TYPE FACE.

ABCDEFGHIJKLMNOPfiRSTUVUXYZ
15345^76^0 YJMSX | »*{>_- + = —

Optional character combinations

j. 2 b a * j i a: i

others available upon request.

♦Previously known as the American Standards Association (ASA). Identified internationally as OCR-A.

Upper case alphabetic USASI characters are located in both shift positions on keyboard 2-900.
Less shifting and greater typing speeds are the advantages of this keyboard arrangement.

\ EDITOR 2

USA STANDARD CHARACTER SET

OCR #2-900 j

d ... 3

Keyboard 2-905 is used with optical readers that recognize different symbols from those on key¬
board 2-900. Both are used for the same applications.

I EDITOR 2

USA STANDARD CHARACTER SET

OCR #2-905|

r -j

102 Composition Input

Keyboard 2-903-1 is used when the OCR typing application contains a large amount of numeri¬
cal information. It contains upper case alphabetic USASI characters and the 10-key keypunch
cluster. Numerics can be entered quickly and easily with this keyboard arrangement.

IBII131IIBIIHUBIIH

IIBIIIBIIBIII

EDITOR 2 USA STANDARD CHARACTER SET

OCR #2 903-1

Keyboard 2-907 is used when typing a large amount of numerics on forms that do not have
pre-printed field separators. This keyboard differs from keyboard 2-903-1 by containing field
separators in both shift positions in place of the group delete. The group delete replaces the
underscore. Less shifting and greater typing speeds are the advantages of this keyboard
arrangement.

MAAG4N 1

1 MAAGIM

&YTASS I

1 « T

IBIIBIII3IIIBIIIBIIIBI

EDITOR 2

USA STANDARD CHARACTER SET

OCR #2 907

Keyboard 2-906 is used with optical readers that recognize different symbols from those on key¬
board 2-903-1. Both are used for the same applications.

TA® I I TA®

CUAft II SCT

EDfTOR 2

USA STANDARD CHARACTER SET

OCR #2 906

Composition Input 103

Keyboard 2-956 contains upper case alphabetic USASI characters and the 10-key keypunch clus¬
ter for rapid entry of numerics. It is recommended only for accounts using competitive typewrit¬
ers to maintain keyboard uniformity.

Keyboard 2-965 contains upper case alphabetic USASI characters and the 10-key keypunch clus
ter for rapid entry of numerics. It is recommended only for accounts using competitive typewrit
ers to maintain keyboard uniformity.

H| bi | ea f mmmmmmmm

1 i TA * 1 V TA# 1

| | CXJLA* || S€T I

iisiliiilisiiisiliiiliailu

!■ IBIIDI |g|||g|[|g|||g|
tel lOlllOIIIBIIIBlIlBI Id

1ISII■ IlSlIlglllHIES

b o | a mmmrn

pa ii hiibhibiiimJ

ASJ151151 ISillSlIli 1

gj[B B M ||tJj|[Og

| EDITOR 2 USA STANDARD CHARACTER SET OCR #2 965j

104 Composition Input

This type style is used for general correspondence
by combining the lower casei illustrated in this
samplei with the USASI type face-

ABCDEFGHIJKLflNOPflRSTUVUXYZ abcdefghijklmnopqrstuvwxyz
15345b7ATD YiPrl** | &* O ?/*

Keyboard 2-901 is used in two ways. The upper case alphabetic USASI characters are used for
OCR typing. For general correspondence typing, use this keyboard in the conventional way.

Composition Input 105

106 Composition Input

8. What a computer does

The Fall of 1962 saw computers utilized in typesetting for the first time.
Both the IBM 1620 and RCA 301 were general-purpose computers
programmed to accept non-justified paper tape input and produce justified
and hyphenated paper tape to activate linecasting units. The perforator
operator typed an “endless” or “idiot” tape with no end-of-line decisions;
however, codes for font, measure, quadding and other typographic functions
were encoded. Essentially the computer performed electronically what the
counting perforator and linecasting spaceband did mechanically. The
computer added the important function of word division. Two methods were
used: logic (or probability) and dictionary (or lookup). Here is an example of
a probability program based on certain rules of logic. This limited but
somewhat effective program is used in Compugraphic phototypesetting
units.

Photo Unit Hyphenation Program

With the hyphen switch in the ON position the following sequence must
occur before a line ending decision is made where the machine will
automatically insert a hyphen:

1. The machine must have read at least two letters in a word.

2. A letter in the Kb group must be followed by a letter in the Kc group:

Kb - B, C, D, F, M, N, P, R, T, V, X, Y

Kc - B, C, D, F, J, M, N, P, S, T, V, W, Z

3. That there are at least 2 letters following the hyphen location to bring
over to the next line.

4. That the word the hyphen location is found in will meet the requirements
to make a line ending decision, ie., that there are enough characters and
bands for justification.

5. That the insertion of the hyphen does not overset the line with band
expansion at minimum.

Discretionary Hyphenation

Discretionary hyphenation requires the operator to insert a tape feed
between letters in a word that is a good hyphenation location. The machine
will insert a hyphen in this position if it is a good line ending point.

Composition Input 107

Dictionary programs stored a large number of words in the computer’s
memory and compared them with words at the line ending point. Correct
break points were indicated. An “exception dictionary” is a limited list of
proper names and other words that do not adhere to the rules of logic. The
computer thus reduces the input burden by eliminating the need for an
operator’s end-of-line decisions.

A general-purpose computer may be programmed to perform a variety of
tasks, such as accounting or billing. A special-purpose computer does only
one job, such as justification. One of the first such special-purpose
computers was the Compugraphic Lineasec, which required a monitor to
make the end-of-line decision for words appearing on a CRT screen. The
next and much more popular version was the “automatic” Justape (also
from Compugraphic). Here, the raw tape entered one side and a justified
tape exited on the other.

Computer programs may be as simple or as complex as the typesetting
device and job to be set require. A machine such as the Photon 532 with 32
typefaces and 23 sizes requires extensive programming to store all the
character width valves and to “mix” typefaces and sizes in the needed
format. A format is a repetitive layout, such as a grocery ad or a book page,
that may be programmed into the computer and accessed via a few
keystrokes. Characters input after a format command are “automatically”
set in the desired face, size and position.

Key to Computerized Composition

Word division and hyphenation are deceptively simple in theory - simply
divide between syllables. But it becomes complex in practice because of
preferential breaks, rule of syllabication and exceptions to these rules, and
exceptions to the exceptions. The collision between the strict logic of
computers and the preferences and exceptions of word divisions has
presented computer programmers with one of their most challenging
problems.

The simplest approach was taken by the Compugraphic Linasec. This
machine relied on a monitor to select hyphenation points when a machine
was unable to end a line on a word space. According to the Linasec
reasoning, many typesetters will not be able to recoup the cost of the large
general purpose computers required to divide words automatically - their
composition volume will not be big enough. Therefore, typesetters need
computing equipment that handles only the specialized work of typesetting,
leaving the work that requires general-purpose capability - hyphenation-to
the best general-purpose computer of all, which is a man. Then came the
mini’s!

Elements other than the word breaks themselves often enter the problem,
such as a case that one programmer encountered. In this instance he told the
computer that a widow of three characters was permissible - whereupon the
computer produced a line containing only a period and two closing quote
marks.

108 Composition Input

This problem, of course, was easily solved by adding extra instructions to the
program defining what was to be considered a character in this instance. But
it lengthened the program to be stored in the computer memory, and
memory capacity is probably the most precious commodity in computers
used for composition. Additional memory capacity can be added to the
system, but only at extra cost.

The choice of program approach often will be dictated by other
circumstances, as the cases of two newspapers illustrate:

At a leading Western newspaper, a logic system based on grammatical rules
of word division was used to guide an RCA 301 computer. It achieves about
85 percent accuracy in hyphenation - that is 85 percent of the lines ending
with a word split are correctly broken according to Webster’s first choice.

Lines incorrectly broken are then reset manually after the type has been cast.
If the computer hyphenates one line in eight, an accuracy of 85 percent will
mean resetting 30 lines in 800 (two lines must be reset for every word division
error), or 3.75 percent of all lines set.

While corrections of this sort are simple to make in hot metal, they are far
more difficult in photocomposed film. A Florida newspaper chain had been
developing film composition techniques for several years, and any computer
installation they made had to be compatible with film composition. They felt
85 percent accuracy would create an uneconomical correction problem.

To increase hyphenation accuracy, they installed an RCA 301 computer
with auxiliary magnetic tape memory capacity. In this memory they stored
some 40,000 root words of five characters and up, in groups to speed access
during operation. Words that break properly according to simple logic rules
are excluded from the computer’s dictionary. With this system, they can
achieve a 99 percent accuracy rate with adequate quality for their
newspapers. This means resetting a maximum of two lines in 800, or a little
more than .3 percent of the total lines produced.

In display advertisements and children’s books lines of text rarely end in a
divided word, since word division would tend to disrupt the reader’s
concentration on the content. Divided words would be unnecessarily
confusing to young readers. Some technical manuals and several newspapers
currently in print also do not contain divided words. In both cases, however,
the reason for not dividing words is primarily an economic one. If the text is
prepared manually, the operator has to make certain hyphenation decisions
which take time and therefore have a cost associated with them. If an
electronic or mechanical device is used for producing text, these devices are
usually less expensive if they do not contain hyphenation capabilities. Thus,
there is definite cost that can be attributed to the division of words, and by
eliminating hyphenation this cost can be reduced or avoided.

The alternative to dividing words at the end of a line is to produce lines of
unequal length by eliminating word division. When these lines are grouped
together into a column of text, the right-hand margin of the column is ragged
rather than flush. Such a column is generally considered harder to read

Composition Input 109

because the eye must constantly adjust its scanning arc. Therefore, a column
with a flush right margin should be easier and faster to read.

The desirability of a flush right margin does not necessarily mean that words
at the end of a line need be divided. There are techniques in current use for
producing flush right margins without resorting to hyphenation. One
technique is to letterspace a word, several words, or a portion of a word in
the line. The extra space that can be used by a hyphenated word at the end of
the line is divided up instead and placed between the letters of a word in the
line. The word or portion of the word is therefore spread apart when it
appears in print. This sometimes makes the letterspaced word extremely
difficult to recognize and impairs speed and comprehension. Another
technique is to take the extra space and insert it in equal amounts between
the words which comprise the line. This produces larger interword spaces,
frequently resulting in distracting and unsightly rivers of white space in a
column of text. Another technique is to divide the extra space on the line by
the number of characters in the line and to increase the set size of each
character by this amount. Varying the set size in effect letterspaces each
word in the line proportionately. Even though the set size of the type may
differ from line to line, there is so little difference that the distraction to the
reader is at a minimum. However, this technique is extremely difficult to
implement.

Also the larger the column measure, the more interword spaces and
characters are available into which the remaining space can be placed,
thereby lessening the need for hyphenation.

Thus, the primary reason for the prevalence of divided words at the end of a
line is basically stylistic. Flush right margins are desirable for both ease and
speed of reading. In addition, the appearance of uniform columns of printed
matter is very pleasing to the eye. In order to achieve this stylistic treatment,
the hyphenation of words at the end of some lines is almost always a
requirement, since the alternatives are graphically unacceptable or incapable
of being produced on the majority of composition devices.

There is an additional reason why hyphenation of text is required. Given a
fixed amount of area into which printed matter can be placed, more text can
be fitted in if words at the ends of lines are hyphenated. Similarly, a given
amount of text can be placed in a smaller area if hyphenation is employed. A
book or newspaper may then require fewer pages, with resultant saving in
paper or composition. This saving can often outweigh the cost of producing
hyphenated text, especially since the inception of computerized text
processing.

Once the requirement to hyphenate words is imposed, the problem becomes
one of simulating the human decision process.

By definition, hyphenation is the division of a word into its component
syllables. A syllable is defined as one or more letters, constituting either an
entire word or part of a word, which are pronounced as a single,
uninterrupted sound. The syllable is the basic unit of pronunciation and
consists of one single prominent sound and usually one or more less

110 Composition Input

prominent sounds. For the majority of English language syllables, the single
prominent sound is usually a vowel, and the less prominent sounds are
usually consonants.

The pronunciation of a word determines the letters comprising each syllable
in the word. A glance at the pronunciation key in the front of any dictionary
shows that the same letter can be pronounced in many different ways. These
pronunciation variations are a problem in any hyphenation routine because
they are not usually reflected in the spelling of the word, and a computer can
use only the alphabetic structure in simulating the human decision process.

A human usually divides the word by pronouncing the word slowly syllable
by syllable. A computer can simulate this process by scanning the word and
isolating the vowels in the word. The isolation of the number of vowels in the
word will not indicate anything about positive hyphen placement; but
hopefully it will indicate the number of syllables in the word. Since the
number of hyphen points in the word is one less than the number of syllables,
this process will yield at least some indication of the number of hyphens to
place in the word. However, this technique is fraught with problems because
the same letter in different words may be pronounced differently or not at
all. For example:

1. The majority of English words ending in the letter E are pronounced so
that the E is silent. The words FACE and WRITE both contain terminal
silent E’s. A computer program that indicates two vowels in each word may
therefore assume incorrectly that each word contains one hyphen point. The
problem cannot be solved by always eliminating a terminal E from the vowel
count because that will produce incorrect results for words like
ASPOSTROPHE, and MAYBE. The problem becomes more complex
when the silent E occurs within a word. In the word BASEMENT, the first E
is not pronounced and therefore should not be considered in a vowel count.

2. A number of English words contain syllables consisting of several vowels
and consonants; for example, TONGUE and TORQUE. In both cases, a
vowel count indicates that several hyphen points can be placed in each word,
whereas pronunciation indicates that the word cannot be hyphenated
because it contains only one syllable.

3. A number of English words also contain consecutive vowels that are
pronounced as one vowel. The words THROUGH and BLEED contain
examples of this condition. A computerized vowel scan can solve this
problem by considering contiguous vowels as a single vowel. However, this
assumption will produce incorrect results in a word like REALITY, where
consecutive vowels are pronounced separately.

4. The letter Y sometimes behaves as a consonant and sometimes as a vowel.
In the word SYLLABLE, the Y behaves as a vowel, while in the word
VINEYARD the Y behaves as a consonant. In general, if a Y is preceded by
a vowel, it can be considered a consonant.

The problems caused by variations in pronunciation are epitomized by the
homographs which exist in the English language. Homographs are words
that are spelled identically but pronounced differently and, therefore,
possibly divided differently. For example, the word PRESENT can be either
a noun or a verb depending upon the context in which it is used. If the word is
used as a noun, it is hyphenated after the S; if it is used as a verb, it is
hyphenated before the S.

Composition Input 111

Each of the problems noted above exists because syllabication is based upon
pronunciation, and pronunciation is often unrelated to the alphabetic
structure of the word. However, there are a number of words which are not
hyphenated according to the way they are pronounced. For example, the
words DOUBLE, MILITIA, and VISION are hyphenated according to
conventions established by printers and writers. Word division is actually
based on many different considerations: pronunciation, conventions and
traditions in general use by printers and writers, etymology, component
elements, context, etc. Considering the diverse bases for word division, it is
understandable why so many inconsistencies exist in the hyphenation of
English words. The inconsistencies resulting from just the variations in
pronunciation should give some idea of the formidable problems that had to
be solved in computerizing word division.

Words which contain alphabetic and nonalphabetic characters or words
which consist entirely of nonalphabetic characters are also problems:

1. Alphameric words consisting of letters and numerals. This type of word
most commonly occurs in parts catalogs and technical publications.

2. Words that contain punctuation marks such as periods, question marks,
and commas, when that word ends a line or a phrase.

3. Compound words or phrases whose elements are separated by text-
supplied hyphens or em dashes. For example, an em dash usually separates
the elements of a telephone number, and a text-supplied hyphen usually
separates the elements of a compound word such as RE-ENTER.

4. Words that contain apostrophes; for example, the possessive form of a
word (HENRY’S) or a contraction (WON’T).

5. Words like “MacDonald” that contain both upper and lowercase
alphabetic letters in the word.

6. Words consisting entirely of nonalphabetic characters, for example,
$12.38 and 1,000,000.

The problem inherent in the above examples is not to recognize the
nonalphabetic characters but to be able to make some decision about
hyphenation-point placement based on these characters. For example, words
which contain text-supplied hyphens can be broken at the point where the
text-supplied hyphen occurs, and words like “MacDonald” can be divided
just before the change from lowercase to uppercase characters. The problem
of dividing mixed words is compounded by the fact that rules for mixed-
word division are determined to a large extent by the stylistic inclinations of
the user. Nevertheless, a generalized hyphenation routine must provide the
ability to hyphenate mixed words.

Not all of the problems associated with computerized hyphenation are due
solely to the idiosyncracies of the English language or to the requirement to
hyphenate mixed words. Two of the primary objectives of any hyphenation
routine must be to maximize hyphenation accuracy and to minimize the time
required to hyphenate the average word. The fastest and most accurate
hyphenation routine might not be practical for the general user if a large and
expensive system configuration were required.

With the application of computers to text processing and the requirement for
112 Composition Input

the division of words at the end of a line, a number of techniques for the
division of words have been developed. In general, these techniques can be
divided into two basic types — dictionary and rules (algorihmic methods.)

Perhaps the simplest computerized word division technique consists of
storing all commonly used words with their associated hyphen points and
then searching this huge dictionary each time a word is to be hyphenated. A
primary disadvantage is that every word requiring hyphenation may not be
in the dictionary. In order for the dictionary to contain all the words that
might need to be hyphenated, especially when one considers proper names
and “made up” words, computer storage requirements and execution time
would be prohibitive.

A second dictionary approach consists of storing commonly used suffixes
and searching this suffix table to see whether the word to be hyphenated
contains one of the suffixes. Since English is fundamentally an agglutinative
language, a hyphen point usually occurs in the word before the suffix. Even
though suffix analysis can pinpoint a word division point in some cases, a
number of exceptions to these rules do exist. For example, the suffix -ING is
usually preceded by a hyphen point, as in the words SAIL-ING, GO-ING,
and COM-ING. However, some technique must be employed to correctly
hyphenate words such as PE-KING, BET-TING, and FLEDG-LING, since
these are exceptions to the general rule. In these cases, the suffixes -KING, -
TING, and -LING can also be made suffixes and placed in a suffix table so
that they will be searched before -ING. Even though a certain amount of
hyphen placement data can be obtained from a suffix analysis, these
examples point out the difficulty of compiling a proper suffix dictionary in
order to minimize the number of exceptions.

A prefix table may also be utilized for the same reasons that a suffix table is
used. But the same problem of exceptions also exists with prefixes. For
example, the prefix TRANS- appears in words like TRANS-PARENT and
TRANS-FORM, but the words TRAN-SCRIPT and TRAN-SCEND
violate the rule. An analysis of a large sample of English language words
indicates that there are a great many more exceptions to prefix analysis than
there are to suffix analysis; for this reason, prefix analysis is difficult to
implement, and quite often the amount of coding needed is not worth the
amount of hyphen placement data obtained.

The second general type of computerized hyphenation technique is generally
referred to as rules or algorithmic. An algorithm is a well defined process or
set of rules for the solution of a problem in a given number of steps. With
relation to hyphenation, the rules are usually the result of extensive
statistical or phonetic and contextual analysis of a large sample of English
words. The development of any algorithm is plagued by two major
difficulties:

1. The accuracy of the algorithm is directly related to the content and size of
the data base from which it is derived. An algorithm based on a small sample
of words which does not include the majority of commonly used words
cannot be expected to correctly hyphenate a large portion of commonly used
words. An algorithm derived from a data base of only scientific terms, for

Composition Input 113

example, may be highly inaccurate when used to hyphenate proper names or
words from another discipline.

2. Normally, most computerized algorithms can analyze a word for division
points based only on the word’s alphabetic structure. Hyphenation of
English words is primarily based on phonetic or contextual considerations;
however, each algorithm must in some way solve this problem if a
reasonable amount of hyphenation accuracy is to be obtained.

Because of the two general difficulties noted above, most of the currently
used algorithms have one major shortcoming —the existence of exceptions
to the rules of the algorithm.

One algorithm that has been used is based on the observation that, for a
greater number of words, hyphen points exist after the third, fifth, and
seventh letters of the word. This algorithm is surprisingly accurate
considering its simplicity, but the number of exceptions to this rule are too
many to permit its use without extensive augmentation with other
techniques.

One algorithmic technique consists of storing, for each of a large number of
strings of characters, the probability that that string of characters comprises
a syllable.

Another algorithmic method that has been developed is commonly known as
the tree technique. Starting at the first vowel of the word, each successive
character in the word is analyzed to see whether this character, in
conjunction with the preceding characters, indicates anything definite about
hyphen placement in the word.

If the character following an O is other than an M, the associated processing
routine may indicate that a third character needs to be analyzed. This
branching process continues until the end of the word is reached, at which
time all the hyphen points in the word will have been found. This technique
is, in effect, a type of dictionary lookup that attempts to pinpoint common
characteristics of character strings without looking at the entire word. If this
technique were flowcharted, the result would be very similar to a tree with its
many branches.

An important algorithmic method of hyphenation is based on the analysis of
the probability that a hyphen point will occur before, between, or after
consecutive two-character combinations comprising the word. Using a large
correctly hyphenated dictionary as the data base, the probabilities are
determined by dividing the total number of times that the two-letter
combinations occur in all the words with a hyphen point between them by
the total number of times that the two-letter combinations occur in the word
regardless of whether a hyphen point separates them or not. The
probabilities themselves are stored in large tables. The algorithm steps
across the word two characters at a time, using the probability for each
possible hyphen point in the word.

A more detailed description of this method may be helpful at this point.
Consider the word PROGRAM. The algorithm initially considers this word

114 Composition Input

to have six positions at which division points can occur:

P-R-O-G-R-A-M
1 2 3 4 5 6

Positions 1, 2, and 6 can be eliminated from consideration, since each
syllable of the word must have at least one vowel in it, and if a hyphen were
placed at any of these points, this condition would be violated. Hence, the
analysis is limited to positions 3, 4, and 5.

Hyphenation accuracy and its effects.

The goal of composition people today seems to be 99 percent accuracy. On
long-measure book work with today’s programs, preliminary experience has
indicated computers will hyphenate between one line in 100 and one line in
1,000. The former hyphenation rate coupled with even 99 percent acceptable
accuracy would require resetting of two lines in 10,000. In a 400 page book,
40 lines per page, this would amount to resetting two to four lines.

The various approaches being taken today seem to be converging toward a
combination system - a basic logic program supplemented by a small
exception dictionary of words that cannot be broken properly by the rules.
The dictionary area of the computer’s memory would also have space
provided for adding or revising words as the language changes.

Opinions differ, however, on how large the exception dictionary will have to
be to achieve this goal. One manufacturer believes it can be reached with
2,000 words, while another maintains it requires 12,000 words. The final
answer depends on several factors, such as the extent to which logic
programs can be perfected and the varying needs of different composing
rooms.

These varying needs are of special significance to book manufacturers today.
Even with computer programmers specifying Webster’s first choice with
their 99 percent accuracy figures, other typographic standards may not be
met adequately by these programs. Most of the work done to date has been
done in newspapers; computer typesetting is only beginning on a large scale
in book manufacturing. And newspaper quality requirements differ greatly
from book requirements.

For instance, newspapers hyphenate proper names, a practice that may not
be permissible in some books. The computer could be given a rule to avoid
this - such as don’t hyphenate words beginning with a capital letter - but this
again takes more space in the computer’s memory as well as slowing down
operations.

Another instance is the breaking of words containing ligatures. Provision for
hyphenating between characters forming ligatures is not included in many
programs currently in use. Programming this requirement is more complex
than a first glance would indicate.

Composition Input 115

FIVE-MINUTE TYPING TEST

Without a doubt very few of us on entering the average modern business
establishment have ever stopped to consider the vital role played by records

in the successful operation of that business. To say that business as we

know it would not be possible without the use of records is certainly no
overstatement.

To say that we live in a world of records is a self-evident truth.

Our lives are bounded by them from the cradle to the grave. The progress of
civilization in the various countries of the world can be measured by the care
and completeness with which the experiences and discoveries of one generation
have been passed on to the next. Business records, naturally, have developed
to the highest degree in centers of commercial activities.

To the uninformed, "Keeping books" may be regarded as an activity of
recent origin and as being responsible for many of the hidden mysteries of
the business world. On the contrary, however, records are almost as old as
history itself. The crude characters scratched on the walls of the caves in
which primitive man existed, the baked clay tablets in the Near East, or the
papyrus rolls in Egypt tell us that since earliest times man has been confronted
with the problems of recording what belongs to him, what is due him, and what
he owes to his creditors in order that he may have some idea of his financial
progress and status. In fact, only two years after Christopher Columbus made
his way westward across the uncharted Atlantic Ocean, the first textbook in
bookkeeping was written by an Italian monk. Those principles of keeping
accounts by the double-entry system as developed in his book are basic to all
our modern accounting records.

Stroke s

148

222

295

311

378

455

534

612

690

750

819

894

970

1047

1124

1205

1283

1361

1439

1515

1588

1667

1697

116 Composition Input

9. Word

processing

This section is included because it will probably require a book of its own
someday. Word Processing is the new name for what we once knew as
automatic typewriting. Almost daily, its application to the composition
function becomes more defined.

If there is any trend that is developing in the composition input area it is
that of “capturing keystrokes.” at the point of origin. The concept is
elementary: at the time someone starts to put an idea on paper capture
some element of that process which can be used as input for typesetting.
The word processor, whatever make, allows ideas to be typed at draft
speeds without regard to errors. Corrections are made only in those areas
that require correction. Thus a typist can produce manuscript copy more
rapidly, and in addition to the typescript maintain some kind of record of
that typing. Magnetic tape cassettes are the most popular form of record
keeping since they can be used over and over again.

The point at which I first write the paragraph is the source documentation
stage. I could be a reporter or an author or anyone who has ideas to
generate and ultimately communicate. I would type this information as
quickly as possible or dictate to a secretary who would do the same. Next
I would go back and change or edit my original material (or the secretary
would). I can change or insert new data because the machine has within it
a flexible recording medium that permits this and the logic to allow it to
be used effectively.

The advantages to anyone who sets type: less redundant keystrokes (don’t
forget that all typesetting material is actually keyed twice, once by the
originator and then by the typesetter), easier correction of author’s
alterations and simple editing. All this is done without a computer. This
chapter will attempt to deal with some of the concepts involved in word
processing and some of the developments that could affect composition
input.

Composition Input 117

Although our ultimate concern is with composition input this area is more
clearly understood by first reviewing the concepts, techniques, and
procedures used to create and prepare material for typesetting.

In essence, one takes an idea and through a series of steps, produces multiple
copies in the form of books, newspapers, or other printed matter.

Creation of material

No matter what the material is, be it a letter, a textbook, a newspaper
article, a magazine, or an advertisement, the starting point is the same: an
idea in the mind of a person. To take this basic idea and create a useful
document or manuscript involves two fundamental functions: the creative
function, and the corrective function.

To start the cycle, an author creates a document containing his ideas. This
may be handwritten; it may be typed in rough draft form; or it may even be
in the form of dictation. The idea then passes to the corrective stage, where
the idea is improved. The form of the improvement can be varied. It may be
a simple spelling correction, it may be improvement of the grammar, or it
may involve massive re-organization of the material with respect to sequence
or content. The improvements are then sent back to the author for his
reaction. The author may massage the material further and then pass it back
to the corrective function for further improvement. It is entirely possible for
material to go through this inter-active “loop” many times before the
material is acceptable. It is also possible for the creative function (author)
and the corrective function (editor) to be the same person. When both
creative and corrective functions agree that the material is acceptable, it is
then considered to be a good manuscript.

Depending upon the remainder of the system, the manuscript may be in one
of two forms: (a) it can be a stack of typewritten pages, or (b) it can be in
machine readable form such as magnetic or paper tape.

Processing of the material

Once the manuscript is informationally correct and in its final form, it
crosses the line from “art” and moves into the field of “craft” where it
begins the printing process.

Typewriting

The creation of a page eventually involves the use of a typewriter to put the
authors’ ideas on paper. Usually this is done at the very start, and the
enrichment cycles involve a re-typing operation to incorporate the changes.
Of course, any typewriter can be used to create correct pages, but our
interest lies with those machines which have been designed to automate and
improve the process.

Step 1

The manuscript, depending upon its form, is given to either a keyboard
operator or a converter operator.

118 Composition Input

a) The keyboard operator simply keys the information onto punched paper
tape. This tape contains both the textual content of the manuscript, plus the
appropriate typesetter control commands for font style, indents, etc.

b) The converter operator loads the machine readable record on the input
side of the converter, and produces a new tape that is capable of driving the
typesetter. This tape contains both textual information, plus appropriate
typesetter commands to control font style, indents, etc.

Step 2

The typesetting process takes the input tape and produces galleys. While the
input tape contains the textual information and some typesetter commands
such as font style and indents, the typesetter is typically adjusted to produce
the desired line length, leading, etc.

Step 3

The make-up step takes the galleys produced by the typesetter and arranges
this information to form a full page. In addition, all artwork, drawings,
illustrations, etc, are added at this step.

Step 4

The full page, called “camera ready copy”, is now passed on to the camera
stage. A photograph is taken of the page, and a negative is made. In full
color printing, a color separation process is used, and the original camera
ready copy is photographed 4 times. A separate negative is made for each of
the 4 colors to be printed, yellow, magenta, cyan, and black.

Step 5

The megative (or negatives) now go to a platemaking process, where
photographically treated aluminum sheet is used to make a contact print of
the negative. In full color printing 4 such aluminum contact prints would be
made, one for each color used. This aluminum sheet is now the offset plate
master.

Step 6

The offset plates are now loaded on the press, and the printing begins. In
black and white printing, the paper need pass through the press only once. In
color printing, the paper needs to pass through the press once for each color,
or more practically, pass through a printing “head” for each color. Thus, full
color offset presses are sometimes referred to as 2 head or 4 head presses.
The output of the press is the finished product, and all that needs to be done
is the appropriate folding, collating, stapling, or binding.

This is one example of the loop that begins with an idea and ends in such a
way as to transmit that idea to others.

Basic automatic typewriters

Virtually every automatic typing system revolves around the use of an

Composition Input 119

electric office typewriter that has been equipped with the facility to accept
electrical signals from the outside to activate typing, and to generate
electrical signals as a by-product of typing. While the form of this capability
varies greatly with manufacturer, these machines are commonly referred to
as I/O (Input/Output) typewriters. The I/O typewriter forms the main
element of the system, and to the typewriter is added a recorder (usually a
slow speed punched paper or magnetic tape device), a reader (again either
magnetic tape or punched paper tape) and the necessary control electronics.
In principle, the typewriter will create a machine readable record as a by¬
product of typing, and it can read this record and produce page copy. Some
systems can read tape, type, and produce new tape simultaneously, while
other systems can only read and type, or type and record.

A number of different typewriters have been equipped with I/O features and
used in automatic typing systems, but the most common machine for this
application is the IBM Selectric, or Golf Ball typewriter. This machine
operates at a printing speed of 15.5 characters per second, and offers the
ability to quickly change character sets and/or styles by simply changing the
ball. A number of type bar machines have also been offered with I/O
features. Type bar machines usually operate at speeds of around 12
characters per second. While there are a number of differences between the
Selectric typewriter and the type bar machines, the two essential ones are
speed and code flexibility. The Selectric is a faster machine, but it has a
rather rigid code structure. The type bar machines, while slower, offer far
greater flexibility for utilizing different code sets.

Typewriter speeds are usually quoted in terms of characters per second,
while typing speeds are quoted in terms of words per minute. Keyboarding
speeds are quoted in characters (or keystrokes) per hour, while Keyboarding
of newspaper material is quoted in terms of lines per hour. The table
furnishes a quick cross reference to these different standards. By definition, a
word is 5 characters plus space, or 6 characters. A newspaper line is
commonly 30 characters.

The concept of power typing

Power typing is a term frequently used by IBM, and in principle is based on
the fact that a girl can type faster when she does not have to worry about
creating error-free page copy. A typical rough draft speed of a typist is 50-60
words per minute when trying to type error-free copy. Power typing involves
giving this girl an automatic typing system that creates a machine readable
record as a by-product of her rough draft typing. Further, the system permits
the girl to back up and overstrike any typing errors she may make, and insert
the correct information. This procedure also backs up the output tape,
deleting the erroneous information and inserting the correct material. It is
important to note that the desired output of this kind of system is a good
tape ; the page produced on the typewriter serves only to give the operator
visibility as she types, and has no further value.

Once the tape has been produced, it can be handled one of two ways.

Stand alone system

In this case, the same girl that prepared the tape in a rough draft typing
120 Composition Input

Electric Typewriters

Adler

Facit

Hermes

IBM

Olivetti

Olympia

Remington

Royal

SCM

Automatic Typewriters

American Automatic Typewriter Co.

Auto-Typist

Metro-tel

Mate

“Automatic” Text Editing Devices
(Stand-alone Work Stations)

IBM

MT/ST-II & IV
MC/ST

Itel

815

852

853

Ty-Data

3600/1

3600/2

3600/3

QuinnData
Quinn Type-70
Quinn Type-80

Edityper

200

Redactron

1 card

2 cards

1 cassette

2 cassettes

Remington Rand

1 card

2 cards

1 cassette

2 cassettes

Wang Laboratories
1210
1220

Multiple Work Station
Dedicated Computer Systems
Information Control Systems Inc.
Astrocomp
Edit Systems Inc.

Text-Ed

Time Shared Text Editing
Bowne Time Sharing
Word I One

Other time sharing companies in U.S.

CRT Stand Alone Devices
Spiras Division of USM Corp.

A ecu text

Lexitron Corporation
Video type

mode now re-inserts this tape into her typing system. The tape now produces
clean page copy at the maximum speed of the typewriter. An example of this
kind of process is as follows: suppose a girl were to type a 10,000 word
document. By using rough draft typing, it will require 250 minutes (at 40
words per minute) to create a good tape of this information. The tape is then
played back through the typing system, and it “power types” at a speed of
155 words per minute (IBM MT/ST), which takes an add minutes. The
entire task has consumed 315 typing minutes, wheras the same task, done on
a standard typewriter by the same girl, would require 667 minutes of typing
time, at an average typing speed of 15 words per minute.

The basic concept of the stand alone system is to use the same machine for
both rough draft typing and power typing.

Centralized typing system

A slightly different set of circumstances surrounds the central typing
application, where there may be many girls transcribing information from a
central dictation source. These systems are commonly used by hospitals and
large insurance offices. The first step in the process is for a girl to transcribe
material in a rough draft typing mode. The good tape that is created by this
process then migrates to a central playback system, where the tapes are
power typed at a speed of 155 words per minute. In this situation, three girls
power typing and one girl operating the playback system can produce the
same amount of finished copy as 7 girls using standard typewriters. The
basic concept of the centralized system is the use of different machines for
each task.

The use of a complete automatic typing system to produce good tapes in the
centralized system is hard to justify, since the desired end result is a good
tape and page is of no value, Opportunities exist for the application of a
lower cost keyboard recorder, which produced good tape as the primary
output but does not deny the operator visibility of her work.

Word processing automatic typewriters

Automatic typewriter systems used for the creating and updating of typed
pages perform the following functions:

- create a machine readable record as a by-product of typing.

- permit re-creation of the correct portions of the copy by using this
machine readable record to operate the typewriter

- permit corrections to be introduced in the proper place in the typed
material, producing new page copy

- create a new machine readable record, containing the useful portions of
the original material plus the additions.

There have been a number of companies that have produced automatic
typing systems, including Facit, Tally, Smith Corona, and others. Of the lot,
IBM, Dura and Friden survive as the largest suppliers of word processing

Composition Input 121

automatic typewriters. These three systems will be exposed in the following
material.

Word processing typing systems are essentially automatic typing systems
with some additional control features provided to simplify and improve the
editing/revision tasks. The required control features are three:

1) Automatic carriage return. This is accomplished by using a “hot zone”
right margin feature. The hot zone is determined by the setting of the right
margin stop, and once the typewriter has typed to the right margin (or hot
zone) a space code or hyphen code causes the carriage to return. This
permits the operator to do some formatting of the typed copy by
determining the line length of the finished page. The machine ignores any
carriage return codes in the input tape, which eliminates the possibility of
format problems due to carriage return codes being discarded with unwanted
material in the input tape.

2) Discretionary and mandatory hyphens. Two kinds of hyphens are used in
word processing typing systems. Discretionary hyphens are those that print
only when they appear in the hot zone. As the original material is typed, the
tape may contain a number of words that are broken by using discrectionary
hyphens. When the page copy is reformatted, these words may not appear at
the end of a line, and so the hyphen is redundant and is not printed.
However, when one of these hyphens appears in the hot zone, it is printed,
and it initiates a carriage return. Mandatory hyphens are those that print
every time, regardless of their place in the line. Some word sequences, such
as “mother-in-law”, require hyphens to always be present. When a
mandatory hyphen appears in the hot zone, it also initiates a carriage return.
The two different hyphens carry two distinct codes; and word processing
typing systems have two hyphen keys on the keyboard.

3) Programmed playback and stop. This feature permits the operator to
select a certain amount of information to type, after which the machine will
stop. Most machines offer some combination of character, word, line,
sentence, or paragraph. In this way, the operator can instruct the machine to
type only a specific portion of the material, which permits additions or
corrections to be inserted in the desired spot.

The logic element controls the operation of the entire system, and this
operation is based on commands from the control panel, or even from
command codes that appear in the input tape. For example, when two
readers are being used in a “tape merge” mode, transfer codes in the input
tapes will shift the input from one reader to the other. Other codes that can
be in the input tape are “punch off’ (where the material is to be typed, but
not copied), etc. The word processor element determines how much material
will be printed before the machine stops (character, word, sentence, etc.).
The example given above uses paper tape terminology, but the same
principles apply to magnetic tape systems.

IBM MT'/ST (Magnetic Tape Selectric Typewriter)

This is the most well known of all the automatic typing systems. The heart of

122 Composition Input

the system is the IBM Selectric typewriter, to which is attached either one or
two magnetic tape stations. Each station is capable of either reading
information into the typewriter or recording material from the typewriter.
The logic in the machine permits the playback to stop after each character,
word, or line. The logic also searches for a given line number in the tape, and
the search speed is 900 characters per second. In addition, the two station
configuration can “merge” tapes, which combines portions of two input
tapes to form a page copy. One of the tapes can serve as a program tape to
determine format, and the other tape serves as the material input tape. The
magnetic cartridges use special edge-sprocketed magnetic tape, very similar
in appearance to 16 mm motion picture film. The cartridges hold up to 100
feet of tape, which will store 24,000 characters, at a price of about $20 each.
Cartridges with less tape are available for $12 each. The cartridges are self
threading, and re-usable many times.

IBM MC/ST (Magnetic Card Selectric Typewriter)

A recent addition to the IBM automatic typing system line is the MC/ST.
While essentially the same as the MT/ST in principle, it differs in one
important aspect. The media used is a special magnetic card, which can store
5000 characters arranged in 50 tracks of 100 characters each. Each card
typically contains a full page of typed information. The MC/ST uses a single
record/replay station, and handles a single card at a time. The card is re¬
usable many times, and cards cost about $1 each.

The significant difference between the MT/ST and the MC/ST is the
concept of updating. The MT/ST reads an input tape, and creates a new
output tape which contains all the valid material from the original tape plus
any additions or corrections. The MC/ST reads the input card, and any
changes or additions are made on that same card, thus in effect updating the
original card. It’s obvious that the MC/ST is therefore limited in the
amount of new information that can be added, while the MT/ST can accept
virtually any amount of new material. The MT/ST is basically an editing
system, while the MC/ST is aimed at the concept of single station power
typing.

Dura

The punched tape equivalent of the MT/ST is the Dura 1041 equipped with
word processor. The heart of the machine is the IBM Selectric I/O
typewriter, to which is connected a slow speed paper tape reader and punch.
The word processor feature permits the playback to occur a word, sentence,
or paragraph at a time. In addition, a single step key furnishes the character
at a time playback. The concept is to read an input tape, insert corrections or
additional information, and create a new output tape as a result. The
machine can also be equipped with an additional reader, which permits
merging of information from two tapes. An additional option permits the
use of a slave printer, which then makes a powerful system for the mass
preparation of personalized letters. The primary input reader contains the
letter text, and the auxiliary reader contains all the names and address. The
main printer types the letter, and the auxiliary printer types the mailing
envelope at the same time the inside address is typed on the letter. In
addition to handling punched paper tape, the machine can be equipped to
handle edge punched cards, either singly or in fan fold strips.

Composition Input 123

The Friden machine has been around for a number of years, having started
out in life as an I/O device for small computers. The heart of the system is
the Friden type bar I/O typewriter, which was originally designed and
manufactured by IBM as their model A typewriter. The typewriter is used in
conjunction with a Friden slow speed paper tape punch and reader, and in
concept is virtually identical to the Dura 1041. The word processer of the
Flexowriter is based on playback of a character, word, or sentence at a time.
Like Dura, the Friden can be equipped with a second reader, which permits
some merging-of input information. The Friden also handles edge punched
cards.

4) Computer managed typing systems

A number of companies are now offering typing systems which address the
very same problem as the word processing systems such as the MT/ST;
however, they utilize a minicomputer as the basic control element. Most of
these systems are designed to furnish a good typed page as the primary
output, and options are available to furnish either paper or magnetic tape as
well. The computer, in addition to controlling the typing of the manuscript
information, can also be used to perform “pre-processing” of the material,
thus organizing the typed material so that the resultant output tape can be
fed directly into a data processing or typesetting system.

A computer managed system is the ASTROTYPE, offered by Information
Control Systems.

The system consists of up to four Selectric I/O typewriters, connected to the
computer. Up to four DEC-TAPE magnetic tape units are connected to the
output of the computer, in effect giving one tape unit for each typewriter,
under computer control. In use, the operator types material at rough draft
speeds (power typing), and the information is captured on magnetic tape.
The computer automatically assigns line numbers to each line as the
operator types. The page copy then passes on to the edit function, where
appropriate notations are made relative to the portions of the text that must
be changed. The page copy comes back to the operator, and now she merely
loads the matching tape, and uses the typewriter to make statements
concerning the action to be taken. The operator types in the number of the
line in question, and then tells the computer what must be done (remove a
word, change a word, change a letter). The computer makes the requested
change, and types out the line for the operator to visually verify. The
operator must acknowledge that the revision is correct, and it then becomes
part of the mag tape record. When all the corrections, additions, or revisions
have been done, the operator calls for a print out of the Final page. The
computer then generates the page copy at the rate of 155 words per minute.
The magnetic tape record of the page is stored until the next revision is
required, or until the tape is erased.

The computer can also do some formatting of the page, such as changing line
lengths, and justifying both margins by interword spacing. The computer can
also remove a paragraph from one section of the page and re-insert it
elsewhere on the page, or it can be inserted on another page.

The system is available in a number of configurations. It can be equipped

124 Composition Input

with a line printer, or it can also be furnished with an output of either
magnetic or punched paper tape. The basic system consists of the computer
(DEC PDP-8), and one typing terminal. To this may be added typing
terminals (up to a maximum of 4 terminals). The system also offers the line
printer.

A somewhat different system is the Varitext, offered by Varian.

The system can be used with up to eight typing stations. The typing station is
an IBM Selectric typewriter, modified by Varian to hold two tape cassettes
of the Philips/Norelco style. In operation, the operator uses the KSR 33
Teletype machine to instruct the system as to the task to be done. Then, the
typing station operators type information in a power typing mode, recording
the information on the cassette in the typewriter. When the page (or pages)
have been completed, the typing station reads this cassette to the computer,
and the computer assigns line numbers and prints the information on a high
speed line printer. The line printer output is then passed to the edit function,
where correctional notations are made relative to the changes required. The
page then returns to the typist, who finds the matching cassette and loads it
in the typewriter, along with a blank cassette. The typist states the line
number in question, and the computer controls the cassettes in the terminal
to copy all the good information from the original cassette to the new one,
stopping at the desired line. The typist enters the correct information, which
is recorded on the second cassette. The typist states the next line, and the
process is repeated until all corrections have been made. When the new
cassette is completed, it can be read to the computer for a new page print out
(either with or without line numbers) or the same cassette can drive the
typing terminal to generate page copy. This system also offers either an IBM
compatible mag tape output, or a high speed paper tape punch.

A still different system is one using a time sharing terminal, such as the IBM
2741 or the Datel. These are typing terminals that can be connected to a
large central computer by means of telephone lines. The concept was
originated by IBM, woo offered their ATS (Administrative Terminal
System for data text) through their Service Bureau Corporation. Several
companies offer versions of the service, among them VIP Systems in
Washington DC.

The typing terminal is an IBM Selectric I/O Typewriter with some
electronics. When not connected to the computer, it operates just like an
office typewriter. In use, the operator connects the terminal to the computer
by daking a telephone call. As information is being typed, it is also being
stored at the computer. The computer automatically numbers lines. Thus,
the page copy is with the operator and the stored information is with the
computer. The page copy is passed to the edit function, and notations
relative to corrections are made. This page then comes back to the operator,
who re-connects the terminal to the computer, types in the in the address of
the information so the computer can retreive it from memory, and then
states the corrections that must be made by using line numbers to pinpoint
areas of interest. The computer makes the corrections, and when all
corrections are complete, the operator can call for a print out of the updated
information. Since the central computer is usually large (such as the IBM
360/65) it is possible for a fully formatted magnetic tape to be made of the
material.

Composition Input 125

Typing terminals are available from a number of sources, and Dura makes a
version of the 1041 typing system for this use. In the case of Dura, the
machine is furnished with a paper tape reader and punch, so that in addition
to the operator having page copy she also retains a paper tape record of the
material.

The rate structure for time sharing varies with the company. In general, the
charges are some combination of the following. The prices shown are those
quoted by one service bureau.

Typing terminal model 2741 $100 per month

Connect time $ 11 per hour

CPU time 150 per second

The CPU time rate is usually governed by the size of the core resident in the
computer. The larger the core size, the higher the rate. In addition, it is
possible to leave information stored with the central computer.

The systems presented in this section have all been based on the principle
that the primary output is a good page ; the good tape is secondary. There
are a few systems (such as the Astrocomp) that are based on the converse,
that is, the good tape is the primary output, the page is secondary. There are
other systems that take this point a step further, and provide a good tape but
no printed output. As a general rule, these systems do not live in both the
world of creation and the world of typesetting.

The typesetting function

The typesetting function embraces the combination of two different types of
information

the text.... what does it say?
the format... how does it look?

Virtually every typesetting device requires that the format control
information be interspersed with the text material. For our purposes, all
typesetting tasks will be placed in one of the three following catagories,
depending upon the amount of format control information required

1) Straight matter or text... doesn’t require much format control

2) Display/advertising material... takes a fair amount of format control

3) Textbooks and special problems ... takes a lot of format control

It’s obvious that the actual procedure in any given case will be determined by
such things as

— the type, and capability of, the typesetting system being used

— the form of the input data

— the output desired

126 Composition Input

Straight matter or text

Straight matter is best illustrated by using a newspaper or paper back book
as an example. The task is to typeset large volumes of material, and the
material is simple in format and uses only a few basic fonts. Straight matter
typesetters usually contain two magazines (each with two rail choices) and
the magazines are loaded with different size fonts, such as 6 point and 10
point. The type styles used are fairly common, such as Bodoni, Garamond,
or News Gothic. The rail choices furnish bold face or italic. The format is
generally consistent, and doesn’t change often. Newspapers use a standard
11 pica line for news columns, with front page or editorial columns going to
15 or 20 pica line lengths. Paper back books are usually around 22 pica line
lengths.

Straight matter tasks can be handled by OCR, or by direct reading of tapes
produced by Dura or Friden or other machines. However, two things must
be done to satisfy the requirements of the typesetting machine:

1) The codes must usually be converted, since computer codes are not
normally compatible with most straight matter typesetters.

2) The textual matter must be combined with typesetter commands to
achieve formatting of the material.

OCR systems usually require two passes at the material before it is ready to
go to the typesetter. The first pass is interpretive, where the pages are read
and converted to machine readable tape, with the output being either 7 or 9
track IBM magnetic tape format, or 8 level punched paper tape.
OCRsystems can read special character sequences which are interpreted as
command codes. Since only a limited number of characters exist on a
typewriter, it is necessary to combine codes to create command sequences.In
this way, the “$” followed by some non-digit such as a letter or a
punctuation mark can be inserted on the page copy to stand for Quad center,
or Upper Rail. During the interpretive pass, this code sequence is simply
converted to a discrete 8 bit code. The second pass is the translative
operation, where the computer compatible tape is read into the system and a
typesetter format tape is generated. Most straight matter typesetters require
6 level TTS tape for input, and in addition, command sequences frequently
take 2 or more codes. The translative pass generates the proper 6 level code
for the textual characters, and “explodes” the 8 level single character into
the appropriate string of 6 level characters for functions. Some of the larger
OCRsystems have sufficient computer capacity to accomplish both the
interpretive and translative functions in a single pass, however the smaller
machines based on the PDP-8 or the Varian generally require two passes.

Classified ads

Newspapers, of course, have requirements in excess of simple straight
matter problems. Such things as classified advertising and display matter are
also routinely done, however in virtually all cases, the tapes to drive the
typesetters are produced on keyboards (as opposed to any OCR type
system). The keyboard produced tapes are non-justified, however there is
seldom any implication of line measure, since any line keyed is less than the

Composition Input 127

line measure being used. For example, most classified ad columns are 9
picas, and provided the material keyed does not exceed 9 picas on any line,
there is no line end decision necessary.

Display typesetting

Newspapers, as well as magazines, use advertising and display copy
throughout their publications. The task is to create eyecatching copy,
arranged in a well balanced aesthetically pleasing manner. This kind of
material combines a variety of different font sizes, font styles, line lenghts,
photographs, and special logos.

Essentially, there are two ways to accomplish this kind of composition.

Paste up

The most common traditional method. In this procedure, all the elements of
the material are typeset separately. For instance, all the 5 point material is
keyed and typeset, then all the 8 point material is keyed and typeset, and so
on. When all the pieces are complete, the ad is assembled in a make up
function. The various textual elements are cut and pasted in their proper
places and at the same time illustrations and photographs are inserted (in the
form of halftones). Large type is usually typeset on a display or headline
machine, such as the CG 7200 or the VGC Phototypositor.

Compose

This procedure involves the use of a phototypesetter (or more properly,
photocomposer) that can typeset a wide variety of font sizes, and styles on a
single piece of film or paper. A number of machines are available today,
starting with expanded machines such as the CG ACM9000 and ending with
the high speed high cost CRT systems. Machines such as the Merganthaler
Super-Quick can hold 8 fonts in sizes from 5 to 72 points. The Photon 713-5
holds 4 fonts, in sizes from 5 to 36 points. Some machines of this class offer
such features as “reverse leading”, where the film or paper can be
“advanced” in either direction to shift the baseline position of the desired
character. Obviously, these additional features greatly complicate the
keyboarding task, since the typesetter control is much more complex. As an
example, a straight matter machine may require 1 or 2 characters to shift
from one font to the other (upper magazine to lower magazine), while a font
change command on a photocomposer may require a code sequence of up to
10 characters. Photocomposers usually require justified tape as an input,
which further complicates the keying task when changing font sizes.

The choice of method (i.e., keyboard and typeset vs paste-up) is usually
made by the mark-up or layout man. To adequately instruct the keyboard
operator he must write down each and every step of the keying process
necessary. Obviously, if the copy to be set is very complex, it may take him
longer to figure out the keyboarding procedure than it would take to
compose the material using cut and paste techniques. On the other hand, he
may know that a certain keyboard operator is very skilled at that particular
kind of keyboarding task, so he will make the decision to have it
keyboarded. The actual decision is subjective, and is influenced by a number

128 Composition Input

of factors that are subject to change, such as the set up of the typesetter at
that moment, the number of skilled keyboard operators available, et al.

In general, there are three ways to create the tape required to drive these
photocomposers:

Counting keyboards

In this method, a counting keyboard is used. The keyboard is adjusted for
the desired line length, and an indicator on the keyboard tells the operator
how much space is left in the line as the keying is done. Where font changes
are necessary, the keyboard operator uses a “look up table” that furnishes
the character sequence necessary to do the task. These counting keyboards
usually have 2 or more width plugs “on line”, to facilitate the use of fonts
with different width values. Counting values can be selected by a single
keystroke. These keyboards are available in either blind or hard copy
versions.

Stored format keyboard

The Automix 710 machine is the primary example of a counting keyboard
with format storage. The 710 is a typewriter based system, which contains 6
programmable “magazines”, and each magazine can be independently
adjusted for a full set of criteria: font size , font style, line length, leading,
space band value, etc.. The operator then uses a single keystroke, calls up the
command sequence and inserts it in the output tape automatically. With this
system, the operator need concentrate only the text, since all the complex
routines are electronically generated.

Computer systems

These systems are extensions of the basic hyphenation-justification
computers used with straight matter machines. The role of the computer has
expanded to embrace photocomposer commands, in addition to the simple
line measure task. In a computer system, the photocomposer control
sequences are resident in memory. At the time of keyboarding, the keyboard
operator would insert an address string of up to 9 characters in the tape.
When this unjustified tape was read into the computer, the address string
would cause the computer to insert the entire photocomposer control
sequence in the resultant output tape. Since the composer command
sequence could be as many as 20 characters, this saved the keyboard
operator a significant number of keystrokes, as well as reducing the
likelihood of errors. These also meant that virtually any keyboard, be it
blind or hard copy, counting or non-counting could be used to prepare input
tapes for computer systems. All that was necessary was the appropriate look
table to inform the keyboard operator of the address string required.

This computer implication did give rise to a family of keyboards that are
specifically tailored for computer input. These keyboards offer an auxilary
keyboard that generates a number of fixed code strings, or addresses, that
can be called by a single keystroke. The code strings are usually factory
wired, and as such are for use with a specific computer, such as the IBM
1130 or the PDP-8 These keyboards are available in either blind or hard
copy versions.

Composition Input 129

Textbooks and special problems

The most complex kind of typesetting usually revolves around the
preparation of textbooks, such as mathematics or languages.

This type of work uses a wide number of special symbols, in addition to both
light and bold face, and italic, typefaces. In addition, material can appear on
different baselines, for both superscript and subscript notation. Typesetters
used in text material composition are usually selected for their ability to

— furnish adequate symbols

— furnish adequate fonts

— furnish adequate sizes of both symbols and fonts

— be able to use multiple baselines

Symbols

The most important aspect. The special symbols are usually grouped
together on “pi” grids on phototypesetters, and on a special “pi” ball for
MT/SC systems. Some phototypesetters furnish standard grids with a
variety of pi characters on the grid. Pi grids are selected for the job at hand,
to address the unique symbols required. For example, mathematics requires
a comprehensive set of special symbols, as does chemistry. Usually the
superscript and subscript characters are furnished on the main grid, or ball,
which reduces the necessity of shifting baselines.

The typesetting of music also requires a large symbol set.

Fonts

Almost any typesetter can furnish adequate fonts for English language work,
however some foreign languages strain the limits of the most sophisticated
typesetters. In addition to requiring a very large number of symbols, often
times the text is set in a different orientation, for example Japanese is set
from the bottom of the page to the top, and Hebrew and Arabic are set right
to left. Arabic is perhaps the most complex of all foreign languages to set,
since not only are there a large number of symbols, but the characters are
linked, similar to handwriting.

Sizes

In general, phototypesetters or composers can easily achieve the desired size
by choosing the right lens. The MT/SC has difficulty with large characters,
and in some cases (such as when printing the mathematical integral symbol)
the MT/SC may print the upper half with one impression, and the lower half
with another impression. Size is probably the easiest problem to overcome
when printing complex material.

Baselines

Most typesetters set material on one baseline at a time, and the paper can be

130 Composition Input

incremented in one direction only. This means that material must be set on
the highest baseline first, starting at the top of the page. The characters that
are to be set appear in the center of the page or on the right, then the carriage
of the phototypesetter must be moved over to the precise spot, the character
set, and then the carriage returned and the paper advanced to the next
baseline. This involves some very complex keying routines when using
phototypesetters, although some phototypesetters offer reverse leading. To
set a superscript symbol, the paper is simply incremented down the required
number of points, and the symbol flashed. Then, the paper is incremented
back to the normal baseline, and setting continues. The reverse procedure is
followed when setting subscripts. The reverse leading feature greatly
simplifies this multiple baseline task.

In practice, the easiest machine to use for complex typesetting is a Selectric
Composer or a Varityper, simply because it has visibility and an operator
can monitor the procedure and make the necessary judgements. By using
Varityper for example, the setting of superscript or subscript is very simple:
when the operator reaches the point requiring the change of a baseline, she
simply rolls the platen up or down the required distance, and sets the
character. This is far easier than programming a phototypesetter to perform
the precise steps necessary. Generally complex material is set by specialty
houses, and they are equipped with all the necessary hardware to do this kind
of typesetting on a production basis. Where smaller commercial printers
have occasion to do this kind of work on a now and then basis, they use
either the Selectric Composer or Varityper, or use cut and paste techniques.

Directory of Word Processing Companies

American Automatic Typewriter Co.

2323 N. Pulaski Rd.

Chicago, Ill. 60639
(312) 384-5151

American Automatic Typewriter Co. is a privately held corporation that has
been manufacturing and marketing “Auto-typist” products for over 30
years. The basic line of products is designed for the production of form or
standard clause correspondence. Direct sales and service organizations exist
in New York City and Chicago, with dealers representing Auto-typist in
most sections of the United States.

Bowne Time Sharing Inc.

345 Hudson St.

New York, N.Y. 10014
(212)989-6006

Bowne Time Sharing Inc. is a subsidiary of Bowne & Co., a large financial
printer. “Word/One,” Bowne’s basic service, is marketed in New York,
Boston, Washington, D.C., Philadelphia and Chicago. Through an

Composition Input 131

arrangement with Pacific International Computer Corp., the service is also
available in San Francisco and Los Angeles. In addition to Word/One, BTS
provides “Photo Comp,” a service that allows Word/One material to be
photo-composed.

Edit Systems Inc.

19959 Vernier Rd.

Harper Woods, Mich. 48225
(313) 886-6545

Edit Systems Inc. provides a shared logic in-house word processing system
called “Text-Ed.” The system utilizes a PDP-8 mini-computer, which
supports Selectric terminals and has a high speed printer. Branch offices are
located in Detroit, Washington D.C., and New York City.

The Edityper Corp. (a Division of Epsco Inc.)

1335 Rockville Pike
Rockville, Md. 20852
(301)424-3997

Epsco Inc. manufactures control systems, data products, and general
electronics. The Edityper Corp. designs and manufactures word processing
equipment. Its manufacturing facilities are located in Westwood, Mass. In
November, 1971, Terminal Equipment Corp. announced agreement with
Epsco to purchase Edityper Corp. Edityper’s products are marketed by
Word Processing Products, Inc., an independent sales and service
organization. WPPI has offices in Philadelphia, Hagerstown, Md. and
Rockville, Md.

Information Control Systems Inc.

313 N. 1st St.

Ann Arbor, Mich.

(313) 761-1600

ICS is in business with a product known as “AstroComp.” AstroComp
utilizes a PDP-8 mini-computer and can support up to 8 Selectric terminals.
A high speed printer is available with the system. Branch offices are located
in Wilton, Conn., and Chicago to serve the New York and Chicago
metropolitan area markets, respectively. Sales representatives also cover St.
Louis, Boston and Washington, D.C.

International Business Machines Corp.

Office Products Division (OPD) Parsons Pond Drive

Franklin Lakes, N.J. 07417

(201)848-1900

There are about 200 OPD sales and service offices throughout the United
States. IBM-OPD provides extensive systems and customer support. The
company also runs a training school (one week) for word processing center
supervisors in Dallas, Tex.

132 Composition Input

Itel Information Products Corp.

(A wholly owned subsidiary of Itel Corp.)

2585 East Bayshore
Palo Alto, Calif. 94303
(415) 328-5660

A network of about 20 branches offices covering 30 cities and about 30
dealer organizations serving 44 cities gives Itel coverage of the major portion
of the U.S. marketplace. Itel’s field force is second only to IBM in word
processing.

Lexitron Corp.

(Formerly Autoscribe Co.)

413 Moss

Burbank, Calif. 91502
(213) 843-51 11

Lexitron builds a uniquely designed CRT display editing device. As yet, the
company has not installed systems, but has announced orders in excess of
$500,000.

Metro-Tel Corp.

409 Railroad Ave.

Westbury, N.Y. 11590
(516) 333-7650

Metro-Tel manufactures telecommunications equipment and automatic
typing equipment. Its Mate automatic typing equipment utilizes a baseplate
that connects to most standard electrics and uses roll paper as the recording
medium. This equipment is used primarily for form letter writing. Metro-Tel
recently introduced Model 801, a magnetic tape cassette unit designed for
more complete word processing applications.

QuinData Inc.

(A division of Quindar Electronics Inc.)

60 Faden Rd.

Springfield, M.J. 07081
(201) 379-7400

Quindar Electronics Inc. is an electronics company that manufactures
advanced solid state equipment for use on data transmission, supervisory
control, computer-based control and/or data acquisition systems. Quin
Data Inc. designs, manufactures, markets and services word processing
products. Manufacturing facilities are in Springfield, N.J. and Toronto,
Ont.

Redactron Corp.

100 Parkway Drive S.

Hauppauge, N.Y. 11787
(516) 543-8700

Redactron was organized in 1969 to develop and manufacture WP
equipment. The company is presently producing both single and dual
magnetic cassette and magnetic card WP devices.

Composition Input 133

Remington Rand
(A division of Sperry Rand)

P.O. Box 1000
Blue Bell, Pa. 19422
(215)646-9000

As a leading manufacturer and marketer of office machines, Remington
Rand has recently completed arrangements with Redactron Corp. to market
their line of cassette and card word processing equipment. It is expected that
Remington Rand will not substantially alter the basic Redactron product at
the outset.

Spiras Systems, Inc.

332 Second Ave.

Waltham, Mass. 02154
(617) 891-7300

Spiras Systems is a subsidiary of USM Corp. Spiras is a producer of mini¬
computers, CRT terminals and a CRT word processing unit known as
“Accutext.” Presently, all of Spiras’ activities are located at headquarters in
Waltham.

Ty-Data Inc.

109 Northeastern Blvd.

P.O. Box 841
Nashua, N.H. 03060
(603) 889-1155

Ty-Data has been manufacturing and distributing its 3600 series of magnetic
tape cassette word processors since early 1971. The dual cassette Model
3600/2 was introduced in October 1971. A three cassette system is expected
by year end. Over 100 units of the one cassette model 3600/1 are in the field.

Wang Laboratories Inc.

836 North St.

Tewksbury, Mass. 01876
(617) 851-7311

Wang is the largest United States manufacturer of electronic calculators. Its
initial WP product entries — the Model 1210 single casette typewriter and
Model 1220 dual cassette typewriter — combine Wang’s expertise in
cassette, typewriter and computer technology. The 1210 and 1220 were
introduced in late October, 1971.

134 Composition Input

The following four pages describe the operating
procedures for the Redactron word processing
system. Much of the information presented is com¬
mon to other systems of its type.

!l!g§gg||glj;

Composition Input 135

GROUP 1.

KEYS DEFINING BASIC OPERATIONS.

Basic operations are something like those in a tape
recorder or dictating equipment. They are indicated
in our Group 1, at the left of the keyboard. We are
talking about them in an order that is logical to the
person using the machine, not the order in which you
see them on the machine.

They indicate the following modes:

a. Record. Original recording of
typewritten information. This
takes place on the magnetic
medium (tape or card) as well
as on paper.

b. Playback. Material stored on mag¬
netic tapes or cards is typed
automatically on paper. "Play¬
back" can be stopped, and ex¬
tra material added to the paper,
then started again. The extra
material on the paper is not
stored, however.

c. Adjust. This is like playback, but
automatically adjusts the right-
hand margin while playing.

d. Duplicate. This is operative on!,
in two-cassette or two-card edg¬
ing typewriters. Information or
one cassette or card is trans¬
ferred to the other at very high
speed, without printing on paper.

e. Edit. This is like playback, but
when printing is stopped, the
operator can type in new ma¬
terial, which is then part of the
magnetic record as well as ap¬
pearing on the paper.

f. Skip/Delete. This, too, is a kinc
of playback. "Skip" is used with
"play" or "adjust" to bypass
material on the tape or card —
that is, not print it. But the
skipped material stays on the
tape or card. When "skip" is
used with "edit", however, ("ed¬
iting" being another word for
correcting), the material is ac¬
tually deleted or removed from
the tape or card.

I RECORD|

BACKUP

WORO END CONT
UNDER UNDER UNDER

REOD MARK

SPACE

GROUP 2.

ACTION KEYS.

They start the tape or card in motion and define
where the action stops. You can think of them also
as positioning keys.

—When the basic operation is "record", the action
keys in the row under the reverse light cause the tape
or card to back up to the specified position.

—When the basic operation is "playback", "adjust",
"edit" or "skip", the tape or card will go forward.

—When "edit" is the basic operation, and the "record"
button is depressed, the keyboard will lock and the
operator may go back by depressing a suitable
action button.

a. Character/stop. The machine
stops at the next character. For¬
ward or reverse.

b. Word. The machine stops after
the next detected space forward.
When backing up for correction
it stops at the first letter of the
word just passed.

c. Line. The machine stops at the
next carrier return in forward.
When backing up, it reverses
one physical block.

d. Find Mark. The machine stops
at the designated "mark". In the
tape machine, the "mark" is the
typist's notation to herself about
a place where she wants a stop
Marks may be "counted" b
the machine so that they can be
returned to by number, at ver,
high speed. Forward or reverse.

In the card machine, "mark
sends you back to the begin¬
ning of the card.

e. Para(graph). The machine (go¬
ing forward only) stops at the
beginning of the next paragraph,

f. Auto. The machine types for¬
ward and stops at the next
"stop" character.

GROUP 3.

KEYS FOR SPECIFIC INSTRUCTIONS
TO THE SYSTEM (CONTROLS).

The keys that are control keys are all regular type¬
writer keys as well. To function as specific instruc¬
tions to the machine, the "Code" key has to be
pushed along with the regular typewriter key. Most of
these keys' additional jobs are indicated on a horizon¬
tal bezel above the keyboard.

Control characters are recorded as single special
characters on the tape or card. If it is desired that they
print, as well, the print control switch (labeled PC) at
lower left is switched on. The character then prints
with a slash on it like this: 2 (The slash of course,
indicates that this is an instruction code, not the
ordinary typing function.)

The control keys are divided into two sets: the keys
we know as format keys (that is,ones we ordinarily
use, like spaces, backspaces, carrier returns); and the
keys we know as number and symbol keys — the top
line of the typewriter.

Taking the formal format keys

first, and from the bottom up:

a. Required space (/). This is a
space that must be considered a
graphic character. It inserts spa¬
ces between words that must be
on the same line, as, for example,
in a date, June 22, 1971.

b. Required carrier return (RCR).
This means that the carrier re¬
turn signal can't be eliminated
in "adjust". This might be used
for the three or four-line ad¬
dress on the top of a letter.

c. Required special carrier return
(SCR). Like the required carrier
return, but the line is not re¬
corded. Used during 1-card, edit,
where it is desirable to get a lot
of information on the card, but
where the width of the line will
be narrower on the paper.

d. Required backspace —). This is
needed to make characters not
on the standard keyboard, as
for example, the accents over e
in some languages.

e. Automatic tab. The typewriter
will automatically execute that
tab after every carrier return.
"Required carrier return" dis¬
continues automatic tab.

Now, taking the character keys

that normally serve graphic func¬
tions:

a. The H is recorded as (£) Mark.
This is used for separating blocks
of material.

c. The (0) is for starting continuous
underlining in groups of words

d. The (9; is for "end underline",
meaning stop the underline.

e. The ($) is for starting word by
word underlining.

f. The (2) is a Stop code, stopping
the machine during playback and
edit when Auto action is used.

g. The (0) is a character called Link
that defines the end of a block
without generating a Carrier Re¬
turn. If it is used with the Special
Carrier Return, it permits cram¬
ming the magnetic medium with
data that will print in the desig¬
nated format.

h. The (8) is an "end of tape" (EOT)
code, for the tape machine. On
the card machine, it means eject
card.

i. The ( 4 ) activates switching from
one card or cassette to the other.
This, of course, is useful only
when there are two. The charac¬
ter may also be called Alternate.
It can be operated in "play" or
"adjust" if you want to manually
switch from one card or cassette
to the other.

j. The (3) is a repeat signal. It will
repeat the Mark previously read
or the entire card.

k. The (2) is for double spacing
material.

l. The (^) is for single spacing
material.

b. The (-) becomes a required hy¬
phen (/) when coded. A required
hyphen is part of a word that
cannot be eliminated if it is run
together (unlike a regular hy¬
phen at the end of a line, but in
the middle of a word). Example:
son-in-law.

GROUP 4.

ADDITIONAL KEYBOARD INDICATORS.

a. At the lower left: the print
control, determines whether the
codes are printed, as well as
recorded.

b. At the upper right: the two keys
designated Units and Tens are
the Mark designating buttons for
the Autofind. These are used to
set the work numbers for the
"Find" counter on the console.
To use them you must depress
the code key. (Tape only.)

c. The light at the right above the
keys is the Back-up light. When
that light is on, back-up direction
is indicated. The buttons immed¬
iately below it cause the tape or
card to go back to the positions
described.

d. The lefthand light on the key¬
board is the Record light, mean¬
ing the tape or card is going
forward. When the light is on,
the information keyboarded is
being recorded.

CONSOLE INDICATORS:

Console — 2 cassette model

ADDITIONAL INDICATORS
ON THE CONSOLE.

The Redactron editing typewriter (single and double card;
single and double tape versions) has a minimum number of
controls on the console.

The card console has an indicator (for track number), and
three buttons: card eject, track up and track down.

The tape version has buttons for: Rewind and Eject,
and the indicator for Autofind. This is an option.

2 6 7
□ □□
4

2 6 7

□ □□

Console — 2 card model

1 Rewind

2 Eject

5 Track Indicators

6 Down

3 Tape Gates 7 Up

4 Card Gates 8 "Find" Counter

Display (2 Digits)

10. Keyboard arrangements

The following pages illustrate some of the keyboard layouts that are
available for composition input. Detailed information on layouts and
operation for specific applications may be obtained directly from the
manufacturers.

Composition Input 139

140 Composition Input

This is the basic layout designed specifically to drive
a Linotype.

JLJ H— LJU
- O' —■'

Z. lu 3

U > 0£L

Q , Qh

DiuDO

O-'OE

Q- Q

O <

f— LU

co q:

a: 3

< <

|_ UJ

CO «

Q a. ■

-l CO

Z Z

LU LU

< <3 ’

* >

> o

VP ® '

t£ 0>

1 oo

* <9
•i N

* a.

% VP

tt in

o o

* <r

cv.cn

5" co
$ °*

THIN

SPACE

CAR

RET

— - <n>

►

z y=

=5 X
CO

H ®

« <l • ^

H—

LL.

h—

z ^

3 X

fc 2

> 3

8 -
l-H

§ u

2 U

<L>

<D J 3
JG *5

4-*

a_, •»

o 1/3
O

3 **■*
O

Cd <D

»—I 4-*

T3 2

cd <u
O X
X> ^
>*

<D cj
44 —

4-»

(A
• 1 "*

• *■*

<D '
Vh ^

Cd g

a, >

£ fc
a>

2 «
43
>» Uh

»—H L

’S C
o 5

5r< O

O o

Vh <m

G to
O W

u &

Vh ~

2 g

S3 g-

u .S

£ CD

13 2

2 cm
2 O

CD M

43 cd
H 43

^ 4-*

C/) Q
<D 4-*

,cd ^

cm or)
<D O

£o

Z £*

G <D

V >

.13 or)
53 5
? *

O Vh
G cd

T3 44
<D

to u

3 u

S3 J

£ CD

m 43
G 4-»

O ^

•<“H Tr

</) O

Composition Input 141

Linotype. This particular layout includes small caps.
There are still quite a few LCCs around, and at used
equipment prices, they can do a satisfactory input

PUNCH TAPE I CODE START STOP STOP LINE I J-CAR

ON FEED I DEL REAO READ CODE DEL I RET

142 Composition Input

The layout utilized a standard typewriter layout with
distinct shift (upper case) and unshift (lower case)
keys. Every line had to end with a “J-Car Return” as
tapes were being prepared.

Composition Input 143

WORD

DEL

(OPT)

QUAD

LEFT

QUAD

CENTER

ADD

THIN

RETURN

PLUS

ELEVATE

x o
O <

c* cm ^

Q O “

c LU
-S CL
2 CL
O =>

<2;

«£in

N^CVI

uj £f

Q uj

O — ,
n lu

u Q

> UJ Q

LU UJ

a < JP

tv.TSl- _J

- o-

c#o)- y

I <x>

If)

H- ^

± u
X O

co —i

m q

CL LU
< LU

D oo
<

Q N“

< 00

<* .
lu zd
CL. <

DC QC ~

UJ o £

Q ^

C£ — J H—

O LU CL.

^QO

LJ LU

x o
0 <
Cm ^

Q O oc

< LJ uj

-S o-
0 3

^ Q-
m CO

Z CL
LU CO

&: cm

w O
O

^ Q -
m co

Z CL

LU CO

<0 00

c* i-*

j vO

•n lO

m CO

? o.
X co

LU “
Q LU

O — 1
u Q

> UJ Q
LU UJ

rv' < UJ

2^
X CO

LU _J

^ <
O 04

Ck CQ

< ~

Q CQ

< hs

s- <

^ <
O £

PAPER FEED

H— O
Q Z
Z <
— X

< i

u O

fV LL

o ^

O Q

a: <

LU LU
M —i

o 9

Q <
< ^

z ^

— M

O CO
CL

LJJ UJ

Q- U

>- <

^ x

U UJ
LU h-

I 00

;£<o

&\n

5 : U
< <
W Q_

Z U

^ 2

UJ Q
CL ill
< LU

g £

O ^

> m
> Q

CO h-
< LU
H— CO

I— <

O cm:

Z ^

— CL

Z P

Q u_

Z “Li

- O

9 z
± <

a ^

Z £

LU O

< I

z “

I I I

0 9

cm <

UJ LU
M —'

a Q
a <
< “J

z ^

^ UJ

LL. Q-

S o

z m

_ M

O CO
Q_

UJ LU
CL U
>■ <

CO ^

o o

U u

uj hH
-7 >
± <
- i co

£ < LU
“■ h- u_

UJ £
Q m

8 =J

u Q

< £

3 uj

0 - 1

Q S

^ <
lu l_:

_ oc

Q LU

< t-

3 Z

0 m
u U

O t

^ "E

3 O

0 DC

o =:

< Q

3 Q

o £

£ O

—w

C'"’G

<* 0 >

I 00

u^cO

^ u
< <
LU CL

z <

^ 2

UJ Q
CL m
< LU
\— LL

rv CO
< ^
5 •"

CO <

< 2
*- 0
LU

9 ^

< LU
I— QC

£i £

3 X

co <0

c* (—

3 “

< z

»— LU

co X

< o

5 Q

< Q

CO £
lu

3 5

< UJ

1 — cm

=! Z

^ “3

CXL

< •“
CJ lli
V on

« ■

43 g

~ o

/—V

«4-< °

«3 C

G *r7

o .s

«o £

i-i *•*

> «-t

£
»-i ^

+■»

•N

eS ^

C/3

43 CD

fli cd
u u

03 W>

43 no

H £

<L>

. O

• 1-4

C" 43

o >>

£ (D

H M

s—✓

V-t

cn a

S ’C

4-*

<D a

43 3
+-» ►—>

a % s

U - £

specific unit. Note that the “Insert Leader” and other
functions are single keystrokes. This keyboard is
made for Compugraphic by Automix.

OH LU

o z

X X

O OJ

£ o

i— cm
Z <

“ £

U _ tz UJ

Eu Z Z x

O O a: 5:

“ £
< t—

x u
u <

i o

z y

LU Z
CO LU

«$N

- iO

£tO

«o-sf

•ft**)

5 cj

LaJ

</>

z u

LU Z
CO LU

•m CT

t; ^

• »H

■M G
CJ

CJ o

•M

M O

Vh +->

/■N #f,H

<2 s

TD Oh
<L> CJ G

M tt) **

goo

° o
* ■■ ^

«»*2 g

CO t« £J
^ c« ^

^ ^ -g

<D G
tj ~ CO

i-j c«

£ u “>

O -*-» G

x *c *-0
JJ £ -

Uh OJ -O

a „

G +-* rj

CJ f\ •

*T-t -M

g 'X -r 1

Lh CO g

e >

„2 5

LU Q
Q- LU
< LU

Q- ^

co (j

laj a.

o_ —

< *
m co

£- Q

O <

h- LU

co cm

E 2
< <
h- ULJ

•- 1 X

. «

•

5>>

-«-»

S-H

•»-*

C/5

3 03

o HH

*t3
cd £

« 12
■S S

co
cd

.2 «>

_S «

p 5 CO

H ss

Composition Input 151

Expanded

Q-

2 r =>
< <

LU zd

Q- <

-♦-*

cd ^
~ cd

.52 a

O (30
<N G
o° -g

e *p

<i> c
O vm

■C °

<L)

<\ Lh

2 w>

fC a}

H *5

c
o

• M
4-*

•

o
a*

S
o
o

X
<L>

a. .

S G
o .2

O CA

CD •—

3 3
O' O

• •

•« <T>

1 CO

^ vo

S in

S. 4*

Z Q

UJ —i

+ 1

—

SPACE

THIN

SPACE

1—

s 9

LU —J

H—

-i

SPACE

VERT

RULE

TAPE

FEED

—J

#• •

SHIFT

h—

LL.

Q Z
< h—

justified cards

The Random Access Composition Entry system which uses punched cards as input to a
phototypesetter.

Memory capacity can be increased by adding discs or tape units.

An editing system. Non-justified input is “massaged” by keyboard and video terminal and
output as justified or unjustified tape.

Wire Service
Lines

Another editing system. Wire Service input is received by a computer and allocated to one
or several video terminals by a controller. After editing is completed, data is output via
punch.

Composition input devices become part of systems.

Composition Input 153

A formatting system. Unjustified data is read into a computer with the capability to
hyphenate using a dictionary and to access a large library of type font widths. Justified
and formatted tapes are then produced to run a particular phototypesetter.

non-justified
mag tape

(or)

justified tape
mag or paper

phototypesetter

accepts non-justified
mag tape

Linecaster or
phototypesetter

A magaetic tape input system.

justified tape mag
or paper

Phototypesetter
accepts non-
justified mag tape

A magnetic tape cassette input system.

154

Composition Input

COMPANY

Alphatype Corporation
7500 McCormick Blvd
Skokie, III 60076
Phone (312) 675-7210

Automix Keyboards, Inc
13256 Northrup Way
Bellevue, Wash 98005
Phone (206) 747-6960

Composition Systems, Inc
325 Central Ave
White Plains, N Y 10606
Phone (914) 761-7800

Compugraphic Corp
80 Industrial Way
Wilmington, Mass 01887
Phone (617) 944-6555

CompuScan, Inc
900 Huyler St
Teterboro, N J 07608
Phone (201) 288-6000

Computer Composition, Inc
16661 Ventura Blvd
Encino, Cal 91316
Phone (213) 986-5552

Datatype Corp
1050 N. W. 163rd Dr
Miami, Fla 33169
Phone (305) 625-8451

Digital Equipment Corp
146 Main St
Maynard, Mass 01754
Phone (617) 897-51 11

ECRM, Inc
1 7 Tudor St
Cambridge, Mass 02139
Phone (617) 661 8600

ETTA Associates
22 Millbrook Lane
Wakefield, Mass 01880
Phone (617) 246-1384

Graphic Products Corp
522 Cottage Grove Pd
Bloomfield, Conn 06002
Phone (203) 243-0730

Graphic Systems, Inc
217 Jackson St
Lowell, Mass 01852
Phone (617) 459-2111

Harris-lntertype Corp
55 Public Sq
Cleveland, Oh 44114
Phone (216) 861-7906

Hendrix Electronics, Inc
Grenier Industrial Village
Londonderry, N H 03053
Phone (603) 669-9050

Imlac Corp
N E Industrial Center
Needham, Mass 02194
Phone (617) 891-1600

Information Control Systems, Inc
313 N. First St
Ann Arbor, Mich 48106
Phone (313) 531-5032

Interface Mechanisms, Inc
5503 232nd St, S. W.

Mountlake Terrace, Wash 98043
Phone (206) 774-3511

Mergenthaler Linotype Co
Mergenthaler Dr
Plainview, N Y 11803
Phone (516) 694-1300

MGD Graphic Systems
3100 S. Central Ave
Chicago, III 60650
Phone (312) 242-4860

Omni-Text, Inc
406 W. Washington
Ann Arbor, Mich 48103
Phone (313) 769-4826

Photon, Inc
355 Middlesex Ave
Wilmington, Mass 01887
Phone (617) 933-7000

Redactron Corp
100 Parkway Dr, S.
Hauppage, N Y 11787
Phone (516) 543-8700

Singer Graphic Systems
2350 Washington Ave
San Leandro, Calif 94557
Phone (415) 357-6800

Star Graphic Systems
2 S. Main St
Hackensack, N J 07606
Phone (201) 652-7200

Tal-Star Computer Systems,
10 Lake Dr

Hightstown, N J 08520
Phone (609) 799-1 1 1 1

Varisystems Corp
80 Skyline Dr
Plainvievv, N Y 11803
Phone (516) 931-7200

VariTyper Division, A-M
11 Mt Pleasant Ave
Hanover, N J 07936
Phone (201) 887-8000

Warlock Computer Corp
Route 7

Georgetown, Conn 06829
Phone (203) 544-8308

Xylogic Systems, Inc
13 Mercer Rd
Natick, Mass 01760
Phone (617) 655-5800

(made by
Graphic Systems)

■M

H—

*03

i—

w w in i/i vi

CD CD 03 03 CD

> > > > >

n <✓> </>

p fl) QJ

C > >

000^5^0000

ccc^^cccc

o o o
c c c

t/3 t/3 t/3 t/3

03 03 03 03

> > > >

$ s g ° g
> > c c

o o o
O if) ld
o_ o CO

co oo" cm"

IT) o
px , o
co , o

in in in tn
0 ) 0)0 0 )
^

cm" cm" cm" cm"

o o o o o

o o © o o

o_ o_ o_ o_ o_

rx' rx' rx" tx rx'

ld id ld id if)

to co co co cd

CO CO CO CO CO

cm" cm" cm" cm" cm"

■|?3

2-^0

! x °
O X o

n. h- CD

>>>>>> c c

CCCCCC 0503

<<<<<<>>

■*-<

1—

■4—*
t/3

O 1

CN 1

“D

0)0 0 ) 0 )
CL Cl Q. Q.

O O o o

4-J 4-» 4-» 4->

D D D D

< < < <

03 0) 0) 0) 0)

1/3 t/3 t/3 t/3 t/3

o o o o o

l. k- i_ L- u.

+-> 4-< +-* +-<

o o o o o

03 03 0) 0) 0)

UJ 11! UJ UJ UJ

o o o o o
o o o o o

o lo o
CD CM O)

oooooooo
oooooococo
co co if) if) if) o

o o o

i—. r—■

o o o o
o o o o

CM CM CM !N

O O O O O

If) If) If) If) ID

O O O ID O O

ID O t- CM O

«- T- CO

o o o o

LO ID O O

t- ID CD

O LD O
»- CM LO

O O O LO LD

ld o id rx rx

o o lo o

LO LO fx LO

0 , 0 ,

O , id ,

■a

r—

*sf

O O ID
O L0 CM
o- CO

OlDOOOOOOO
IXLOCMOOlOCMOLD
«- >- CO CO 00

o o o o o o

CO CO o- CO

o o o o

O O LO LD

o o o o

O O O LD
CO CM o- «-

<—

f^.

r>x

r^-

r>.

p*.

r^.

r-x

r*x

0 )

®ooo*-»-

co rx rx ix rx rx

O) 0 ) O O O) o

O O O r-
rx ix rx rx
0 0)0)0)

Q O ^ r- CM

cd rx rx rx rx

o o o o ©

OOOOOOOOOOO
LOLOLOOLOLOLOLOlOlOLO
0) C) 0) D G) O) G) O) O) C>
|x" |x rx O’" O O’ O’ CN O CD «—

O O O

o o o

ODD

D O’ <—
CD D i-

O O O O O O O -vT fo -
OOODOOO^SS
o-oocmooo c ? ic ) j
d" cm" cd" <-' d" d" 2 2

CDO-DCMO-CDCD22

o o
o o

D D

cm" cd" 9
™ ™ $
° ° co"

o o o o o o

D D D O O O

O O O D D D

o o o o
o o o o

D O CM D

D CD CD O O’ CD CM N O’ CM

CM CM CM i- <- r- CM

O O O D O O

O D D CO D O

LD_ 0) O) 00 0) LO_

co" rj" r-'" jx."

CM «- r- CM «-

t/3 t/3 t/3 O t/3

03 03 03 r- 03

> > > >

o o o
c c c

o <33
C 03 03

1/3 t/3 CO C/3 1/3 t/3 O

03 03 03 03 0 03 r-

>>>>>>

t/3 t/3 t/3

03 03 03

> > >

t/3 t/3 t/3 t/3

03 03 03 03

> > > >

t/3 t/3 t/3 t/3 t/3

03 03 03 03 03

> > > > >

cd cd cd

“0

•0

■0

-Q

+-»

4-J

4—*

4->

4—

CD CD CO CO CO
q * * ' * CO -

_q co co r^.

S' cd" cd" cd id id . id

|x CD CD CD CD

CO CD CD CO

co oo co oo oo _5

i*'* rx' rx" rx' rx" >

tO CD CO CD CD

S I 8

J 0) Q.

CO LD LO CO LD
CO -*fr CO

LD LD CD
^

^ CM CN

tf^COLDLDLDLD^'sf
LOLD^^-'sJ'^^LDlD
LD LO

LD LD LD
^ ^

CO CO CO LD
CO CO CO ^

LD CO CO LD LD

^ co co

CN^^tCN^CNCNCNCN^

^(NCMr-cNrNr-rsrsCN V

LDLDLDLDLD^-O'Sj'CDCD
r- CO t-

CN CN

CNCNCNCDCDCDCDCNCN^-

r^r^r^cocococor^r^^

CDCDCDCDCDCDCOCDCDCO

CM CN CD CD CD CD
r- r- C[) CO CO

LD LO CN LD LD LD

^ co

CM CN CM
LD LD ID ID
LD LD LD

CO CO CO CO CM ^

T-T-T-T-r^T-

CD CD ID LD LD CD

r-r-r-r-CNCOCOCOCN

CMCM^COCOCOCO^^CO
T- T- CN T T ^

r- *— Q O LD LD

O) CN ^ O CO
CM

OOOOOCMCNCNCNOO

0)0^0)0)0)00000)03

O O O
CN CM CM

OOOCDCDCDCDCNCMO
0)0)0)0)030)0)'— *— O

O ^ CN CM CN
03 + 0 0 0

O O O
0)0*-

CN O CN CN CN O
*- O O

CNCNCNCN^^^^COCM

LD T CD
*- O

CDCDCN'tOOCO’tfCOOO^

^ <3*

CN CN ^ ^ ^

CN CM CN CD

CO CN ^ CO ^

O LD LD O LD
CD CN CM CO CN

O LD o
CO *7 LD
6
o

ldldooooo|£

CN CN tt O O LO ',
"—CM O
®

If) If) Lf)

CO 00 CO

o o o o

if) rf

o o o o o
cm <a- in oo if)

.% 2 g 8 I

CO —I —I —I'D .E o §

XH-H b- X q O) q.

t— t— t— 0*”00 -m^ c

cDCDr^r^cDOO of; £
OOOOJOCMCMI^O
CMCMCNCN^fr^r^Q-<lCJ

o u-
° C o

CN >< O
r- h (O

c c c
o o o

o o o

UL LL LL

o o
o o o

LT) r- r- CN
CN CO CO CO CO

O'— CN CO CO CO CO u o %

O CD CO '— '— '— '— CD CD ^

CNLDLor^r^r^r^o-Q-X

•D o O O
^ o o o

> O *- CM

ro oo co a)

CL x X X

X X 'P 'P 'P

03 CD “o c C C

^ ^ o o o o

ts « O o o o

0 d r r r r

-> -3 CL Q_ Q_ CL

D _J

x x Q

O O O r-

ro ® ® ®

in rx n- rx co c

CMO'f'^M’i-
cx rx. rx. ^

2^225 S

< < < < < X

156 Composition Input

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Texts

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "Condensed Handbook Of Composition Input (book)"

See other formats